{"publisher":"Anawords","url":"https://anawords.com/","labels":{"confirmed":{"name":"Confirmed","meaning":"An official source says it: lab docs, pricing page, notice."},"reported":{"name":"Reported","meaning":"Two or more independent, named outlets say it."},"claim":{"name":"Company claim","meaning":"Only the company says it. Not checked by anyone else yet."},"unconfirmed":{"name":"Unconfirmed","meaning":"One report only, shown with its source."},"unknown":{"name":"Not known","meaning":"Not announced or not answered yet."}},"stories":[{"url":"https://anawords.com/story/asml-q3-2026-results-ai-demand-capacity-2028","title":"ASML says AI demand means its 2027 and 2028 supply will fall short, after record €11.7 billion quarter","summary":"ASML reported €11.7 billion in Q3 2026 sales and €5.0 billion net income, guided to €44–45 billion for 2026 and said 2027–28 supply will not meet AI-driven…","section":"Chips & Compute","published":"2026-10-15T12:24:51+05:30","updated":"2026-10-15T12:24:51+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/asml-q3-2026-results-ai-demand-capacity-2028","facts":[{"text":"ASML’s Q3 2026 total net sales were €11.7 billion and net income €5.0 billion.","label":"Confirmed","sources":[1,3]},{"text":"ASML expects 2026 sales of €44–45 billion.","label":"Confirmed","sources":[1]},{"text":"ASML says planned 2027 and 2028 supply will not fully meet customer demand.","label":"Confirmed","sources":[1,3]},{"text":"Capacity for low-NA EUV and DUV immersion will rise 30% in 2027.","label":"Confirmed","sources":[1]},{"text":"China’s share of system sales rose to 23% from 14%.","label":"Unconfirmed","sources":[2]},{"text":"Financial guidance for 2027.","label":"Not known","sources":[]}],"sources":[{"title":"ASML Q3 2026 results press release (SEC Form 6-K)","url":"https://www.sec.gov/Archives/edgar/data/0000937966/000162828026065813/pressreleasefinancialresul.htm","type":"official"},{"title":"South China Morning Post: ASML’s chip equipment sales to China soar 109%","url":"https://www.scmp.com/tech/big-tech/article/3370904/asmls-chip-equipment-sales-china-soar-109-despite-us-export-curb-push","type":"outlet"},{"title":"Investing.com: ASML raises capacity plans as AI demand drives record Q3 sales","url":"https://uk.investing.com/news/earnings/asml-raises-capacity-plans-as-ai-demand-drives-record-q3-sales-4908944","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/asml-q3-2026-results-ai-demand-capacity-2028.json"},{"url":"https://anawords.com/story/anthropic-scion-source-control-replaces-git-monorepo","title":"Anthropic says it replaced Git with its own source control system, Scion, after daily merges passed 30,000","summary":"Anthropic described Scion, an in-house version control system that replaced Git for its monorepo. Its numbers, what is released, and the Mononoke credit…","section":"Tools & Apps","published":"2026-10-15T12:20:51+05:30","updated":"2026-10-15T12:20:51+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/anthropic-scion-source-control-replaces-git-monorepo","facts":[{"text":"Anthropic moved its monorepo from Git to its in-house system, Scion, in five months.","label":"Confirmed","sources":[1]},{"text":"Daily merges grew from about 2,500 in May 2026 to over 30,000 in September.","label":"Company claim","sources":[1]},{"text":"Scion’s infrastructure is 200x smaller than the old Git mirrors.","label":"Company claim","sources":[1]},{"text":"Scion is internal and has not been released as open source.","label":"Confirmed","sources":[1]},{"text":"A former Mononoke developer says Scion’s design closely follows Mononoke and should credit it.","label":"Confirmed","sources":[2]},{"text":"Whether Anthropic will respond to the credit criticism.","label":"Not known","sources":[]}],"sources":[{"title":"Anthropic Engineering: Redesigning source control","url":"https://www.anthropic.com/engineering/redesigning-source-control","type":"official"},{"title":"sunshowers: Anthropic’s Scion and Mononoke","url":"https://sunshowers.io/posts/anthropic-mononoke/","type":"outlet"},{"title":"AI Weekly: Anthropic builds Scion to replace Git","url":"https://aiweekly.co/alerts/anthropic-builds-scion-to-replace-git-after-merges-hit-30000day","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/anthropic-scion-source-control-replaces-git-monorepo.json"},{"url":"https://anawords.com/story/xai-grok-voice-3-speech-to-speech-model","title":"xAI releases Grok Voice 3.0 for voice agents at the same $0.08-a-minute price as before","summary":"xAI released Grok Voice 3.0 on 14 October 2026 in its Voice Agent API at $0.08 a minute. What it does, which scores are xAI’s own, and where it trails rivals.","section":"Models & Launches","published":"2026-10-15T12:16:51+05:30","updated":"2026-10-15T12:16:51+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/xai-grok-voice-3-speech-to-speech-model","facts":[{"text":"xAI released Grok Voice 3.0 on 14 October 2026 in its Voice Agent API.","label":"Confirmed","sources":[1]},{"text":"It costs $0.08 a minute, the same as Grok Voice 2.0.","label":"Confirmed","sources":[1,2]},{"text":"It scores 86.9 on the Artificial Analysis Speech-to-Speech Index, ahead of GPT-Live-1 at 83.2.","label":"Company claim","sources":[1,2]},{"text":"On τ-Voice it scores 71.6%, behind GPT-Live-1 at 74.5%.","label":"Company claim","sources":[1]},{"text":"Starlink resolves about 70% of some 20,000 daily calls without a person.","label":"Company claim","sources":[1,2]},{"text":"Supported languages and voices.","label":"Not known","sources":[]}],"sources":[{"title":"xAI: Grok Voice 3.0","url":"https://x.ai/news/grok-voice-3","type":"official"},{"title":"AI Weekly: xAI Grok Voice 3.0 ships at $0.08/min","url":"https://aiweekly.co/alerts/xai-grok-voice-30-ships-at-008min-tops-index-at-869","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/xai-grok-voice-3-speech-to-speech-model.json"},{"url":"https://anawords.com/story/eight-states-sue-openai-chatgpt-child-safety","title":"Eight US states sue OpenAI, saying ChatGPT harmed children and that the company misled the public about its safety","summary":"Eight US state attorneys general sued OpenAI on 14 October 2026 over ChatGPT’s effects on minors. Which states, which laws, what they ask for and OpenAI’s…","section":"Policy & Regulation","published":"2026-10-15T12:12:51+05:30","updated":"2026-10-15T12:12:51+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/eight-states-sue-openai-chatgpt-child-safety","facts":[{"text":"Eight state attorneys general filed separate lawsuits against OpenAI on 14 October 2026.","label":"Reported","sources":[1,4]},{"text":"Arizona sued in Maricopa County Superior Court under its Consumer Fraud Act.","label":"Confirmed","sources":[2]},{"text":"Montana sued under its Consumer Protection Act and Consumer Data Privacy Act.","label":"Confirmed","sources":[3]},{"text":"The states allege ChatGPT harmed minors and that OpenAI misled the public about its safety.","label":"Confirmed","sources":[2,3,4]},{"text":"OpenAI says it has built age-specific protections and will add more safeguards.","label":"Company claim","sources":[1]},{"text":"When the cases will be heard.","label":"Not known","sources":[]}],"sources":[{"title":"Bloomberg Law: States hit OpenAI with kids’ online safety lawsuits","url":"https://news.bloomberglaw.com/privacy-and-data-security/half-dozen-states-hit-openai-with-kids-online-safety-lawsuits","type":"outlet"},{"title":"Arizona Attorney General: Lawsuit against OpenAI","url":"https://www.azag.gov/press-release/attorney-general-mayes-announces-lawsuit-against-openai-deceiving-arizonans-about","type":"official"},{"title":"Montana Department of Justice: Knudsen files lawsuit against OpenAI","url":"https://dojmt.gov/attorney-general-knudsen-files-lawsuit-against-openai-for-endangering-children-and-teens/","type":"official"},{"title":"KWCH: Kansas AG Kris Kobach announces lawsuit against OpenAI","url":"https://www.kwch.com/2026/10/14/kansas-ag-kris-kobach-announces-lawsuit-against-openai/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/eight-states-sue-openai-chatgpt-child-safety.json"},{"url":"https://anawords.com/story/openai-gpt-4o-gpt-4-1-api-shutdown-january-2027","title":"OpenAI will remove GPT-4o and GPT-4.1 from its API on 14 January 2027 and points developers to GPT-6 Luna","summary":"OpenAI will remove GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini from its API on 14 January 2027. Which versions, the recommended replacement and its price.","section":"Models & Launches","published":"2026-10-15T11:55:49+05:30","updated":"2026-10-15T11:55:49+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/openai-gpt-4o-gpt-4-1-api-shutdown-january-2027","facts":[{"text":"GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini versions will be removed from the API on 14 January 2027.","label":"Confirmed","sources":[1]},{"text":"Fine-tuned versions follow the same shutdown date as their base models.","label":"Confirmed","sources":[1]},{"text":"OpenAI recommends gpt-6-luna as the replacement for all four.","label":"Confirmed","sources":[1]},{"text":"gpt-6-luna costs $0.10 input, $0.01 cached input and $0.50 output per million tokens (standard tier).","label":"Confirmed","sources":[2]},{"text":"GPT-4o and GPT-4.1 were retired from ChatGPT on 13 February 2026.","label":"Confirmed","sources":[4]}],"sources":[{"title":"OpenAI API: Deprecations","url":"https://developers.openai.com/api/docs/deprecations","type":"official"},{"title":"OpenAI API: Pricing","url":"https://developers.openai.com/api/docs/pricing","type":"official"},{"title":"Shelly Palmer: OpenAI will sunset selected GPT-4 models in January","url":"https://shellypalmer.com/2026/10/openai-will-sunset-selected-gpt-4-models-in-january/","type":"outlet"},{"title":"OpenAI: Retiring GPT-4o and older models","url":"https://openai.com/index/retiring-gpt-4o-and-older-models","type":"official"}],"model_card":null,"data":"https://anawords.com/story/openai-gpt-4o-gpt-4-1-api-shutdown-january-2027.json"},{"url":"https://anawords.com/story/google-gemini-4-argon-available-developers-gemini-app","title":"Google opens Gemini 4 Argon to developers worldwide and to paid Gemini subscribers","summary":"Gemini 4 Argon is now in the Gemini API worldwide and in the Gemini app and AI Mode for AI Pro and Ultra users. Prices until 30 November, benchmarks and…","section":"Models & Launches","published":"2026-10-14T16:01:37+05:30","updated":"2026-10-14T16:01:37+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/google-gemini-4-argon-available-developers-gemini-app","facts":[{"text":"Gemini 4 Argon is available worldwide via the Gemini API from 13 October 2026.","label":"Confirmed","sources":[1,2]},{"text":"AI Pro and Ultra subscribers can use it in the Gemini app and AI Mode.","label":"Confirmed","sources":[1,2]},{"text":"Introductory prices are $2 and $10 per million tokens until 30 November, then $4 and $20.","label":"Confirmed","sources":[1,2]},{"text":"First on the Vals Index; 87% on ERQA.","label":"Company claim","sources":[1]},{"text":"Matches GPT-6 Astra on Artificial Analysis’s Intelligence Index.","label":"Unconfirmed","sources":[2]}],"sources":[{"title":"Google: Gemini 4 Argon is now available for everyone","url":"https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/","type":"official"},{"title":"Engadget: Google’s Gemini 4 Argon is now available in the Gemini app and AI Mode","url":"https://www.engadget.com/2285023/google-gemini-4-argon-is-now-available-in-the-gemini-app-and-ai-mode-in-search/","type":"outlet"}],"model_card":{"name":"Gemini 4 Argon","lab":"Google DeepMind","released":"13 Oct 2026 (general)","context":"1M tokens (shared input/output)","input_price":"$2 (intro), then $4","output_price":"$10 (intro), then $20","india":"API global; app needs AI Pro/Ultra","licence":"Proprietary","replaces":"Not stated"},"data":"https://anawords.com/story/google-gemini-4-argon-available-developers-gemini-app.json"},{"url":"https://anawords.com/story/google-play-personal-vibe-coded-apps","title":"Google Play lets people run their own AI-built “personal apps” on Android without publishing them","summary":"Google Play now lets US adults run private, AI-built “personal apps” made with Google AI Studio, visible only to them. How it works, limits and what comes…","section":"Tools & Apps","published":"2026-10-14T15:56:37+05:30","updated":"2026-10-14T15:56:37+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/google-play-personal-vibe-coded-apps","facts":[{"text":"People can run private AI-built apps through Google Play, visible only to them.","label":"Unconfirmed","sources":[1]},{"text":"It starts for users 18 and over in the US.","label":"Unconfirmed","sources":[1]},{"text":"Personal apps must follow Play policies; public release needs trust and safety review.","label":"Unconfirmed","sources":[1]},{"text":"Google AI Studio has built native Android apps from prompts since May.","label":"Unconfirmed","sources":[2]},{"text":"When it reaches India.","label":"Not known","sources":[]}],"sources":[{"title":"TechCrunch: Google Play now lets anyone build private, vibe-coded apps","url":"https://techcrunch.com/2026/10/13/google-play-now-supports-private-vibe-coded-apps-built-just-for-you/","type":"outlet"},{"title":"TechCrunch (May 2026): Google’s AI Studio now lets anyone build Android apps in minutes","url":"https://techcrunch.com/2026/05/19/googles-ai-studio-now-lets-anyone-build-android-apps-in-minutes/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/google-play-personal-vibe-coded-apps.json"},{"url":"https://anawords.com/story/x-subscription-grok-cursor-x-premium-bundle","title":"SpaceX’s AI business bundles Grok, Cursor and X Premium into one subscription from $8 a month","summary":"The new X Subscription bundles Grok, Cursor coding agents and X Premium in plans from free to $200 a month. What each tier includes and how existing users…","section":"Tools & Apps","published":"2026-10-14T15:51:37+05:30","updated":"2026-10-14T15:51:37+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/x-subscription-grok-cursor-x-premium-bundle","facts":[{"text":"The X Subscription launched on 12 October 2026, bundling Grok, Cursor and X Premium.","label":"Confirmed","sources":[1,2]},{"text":"Plans are Free, $8, $30, $100 and $200 a month.","label":"Confirmed","sources":[1,2]},{"text":"Premium includes SuperGrok, Cursor Pro and X Premium+ benefits.","label":"Confirmed","sources":[1]},{"text":"Existing subscribers can switch with a prorated refund.","label":"Confirmed","sources":[1,2]},{"text":"India pricing.","label":"Not known","sources":[]}],"sources":[{"title":"xAI: The X Subscription","url":"https://x.ai/news/x-subscription","type":"official"},{"title":"TechCrunch: SpaceXSI bundles X Premium, Grok and Cursor into one subscription","url":"https://techcrunch.com/2026/10/13/spacexsi-bundles-x-premium-grok-and-cursor-into-one-subscription/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/x-subscription-grok-cursor-x-premium-bundle.json"},{"url":"https://anawords.com/story/harvard-study-ai-coding-agents-more-code-not-more-software","title":"AI coding agents add 30% more code but not more finished software, Harvard researchers find","summary":"Harvard economists studied 700+ firms: AI coding agents raised lines of code about 30%, but finished tasks did not rise and reviews slowed 49%. Findings and…","section":"Research","published":"2026-10-14T15:46:37+05:30","updated":"2026-10-14T15:46:37+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/harvard-study-ai-coding-agents-more-code-not-more-software","facts":[{"text":"Lines of code per developer rose about 30% after AI coding agents arrived.","label":"Reported","sources":[1,2]},{"text":"Completed tasks showed no statistically significant change.","label":"Reported","sources":[1,2]},{"text":"Time from submitting code to merging rose 49%.","label":"Reported","sources":[1,2]},{"text":"The study has not been peer reviewed.","label":"Unconfirmed","sources":[2]},{"text":"Whether AI agents produce more faulty code.","label":"Not known","sources":[]}],"sources":[{"title":"AI Weekly: Harvard study: AI coding agents lift code 30%, shipped output flat","url":"https://aiweekly.co/alerts/harvard-study-ai-coding-agents-lift-code-30-shipped-output-flat","type":"outlet"},{"title":"Netzwoche: Harvard study questions AI productivity promise","url":"https://www.netzwoche.ch/news/2026-10-12/harvard-studie-stellt-produktivitaetsversprechen-von-ki-infrage","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/harvard-study-ai-coding-agents-more-code-not-more-software.json"},{"url":"https://anawords.com/story/meta-github-breach-stolen-credentials-1500-notified","title":"Meta says attackers used an employee’s stolen login to copy code from some of its GitHub repositories","summary":"Meta says outsiders used an employee’s stolen credentials to access some GitHub repositories, taking mostly code and some personal data of about 1,500…","section":"Safety & Trust","published":"2026-10-14T15:42:09+05:30","updated":"2026-10-14T15:42:09+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/meta-github-breach-stolen-credentials-1500-notified","facts":[{"text":"Attackers used an employee’s stolen credentials to access some Meta GitHub repositories.","label":"Reported","sources":[1,2]},{"text":"About 1,500 people, mostly US employees, are being notified.","label":"Reported","sources":[1,2]},{"text":"99% of the data taken was code; 1% was personal data.","label":"Company claim","sources":[1,2]},{"text":"Meta says no product was made vulnerable.","label":"Company claim","sources":[1]},{"text":"Who the attackers were.","label":"Not known","sources":[]}],"sources":[{"title":"SecurityWeek: Meta says hackers took code, personal data of 1,500 in GitHub breach","url":"https://www.securityweek.com/meta-says-hackers-took-code-personal-data-of-1500-in-github-breach/","type":"outlet"},{"title":"AI Weekly: Meta says hackers stole code and personal data of 1,500 employees","url":"https://aiweekly.co/alerts/meta-says-hackers-stole-code-and-personal-data-of-1500-mostly-us-employees-via","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/meta-github-breach-stolen-credentials-1500-notified.json"},{"url":"https://anawords.com/story/apple-tv-4k-a19-siri-ai-far-field-microphones","title":"Apple’s new Apple TV 4K brings Siri AI with hands-free voice control, on sale 13 November","summary":"Apple’s new Apple TV 4K has the A19 chip, Siri AI and three far-field microphones for hands-free control. $299, pre-orders open, available 13 November…","section":"Tools & Apps","published":"2026-10-14T13:39:49+05:30","updated":"2026-10-14T13:39:49+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/apple-tv-4k-a19-siri-ai-far-field-microphones","facts":[{"text":"The new Apple TV 4K has the A19 chip and three far-field microphones.","label":"Confirmed","sources":[1,2]},{"text":"It is available from 13 November, with pre-orders open.","label":"Confirmed","sources":[1,2]},{"text":"It is priced at $299.","label":"Unconfirmed","sources":[2]},{"text":"Siri AI is rolling out in English and usage limits may apply.","label":"Confirmed","sources":[1]},{"text":"India price and availability.","label":"Not known","sources":[]}],"sources":[{"title":"Apple: Apple TV 4K","url":"https://www.apple.com/apple-tv-4k/","type":"official"},{"title":"TechCrunch: Apple launches a new Apple TV 4K with Siri AI","url":"https://techcrunch.com/2026/10/13/apple-launches-a-new-apple-tv-4k-with-siri-ai-and-remote-free-commands/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/apple-tv-4k-a19-siri-ai-far-field-microphones.json"},{"url":"https://anawords.com/story/apple-openai-trade-secrets-expedited-discovery-ruling","title":"Judge orders OpenAI to hand Apple some device forensics early in Apple’s trade secrets lawsuit","summary":"A US judge partly granted Apple’s request for early evidence in its trade secrets suit against OpenAI and former Apple staff. What was ordered, denied and…","section":"Policy & Regulation","published":"2026-10-14T13:34:49+05:30","updated":"2026-10-14T15:42:09+05:30","version":2,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/apple-openai-trade-secrets-expedited-discovery-ruling","facts":[{"text":"The judge partly granted Apple’s request for expedited discovery on 13 October 2026.","label":"Reported","sources":[1,2,3]},{"text":"OpenAI must produce device forensic images it has already collected.","label":"Reported","sources":[1,2]},{"text":"Discovery on metal finishes, power management and batteries was denied for now.","label":"Reported","sources":[1,2,3]},{"text":"OpenAI says the lawsuit is baseless.","label":"Reported","sources":[1,2]},{"text":"Whether any trade secrets were taken.","label":"Not known","sources":[]},{"text":"OpenAI has five days for collected materials and 14 days for others.","label":"Unconfirmed","sources":[3]}],"sources":[{"title":"Bloomberg Law: Apple wins early access to OpenAI files in trade secrets suit","url":"https://news.bloomberglaw.com/tech-and-telecom-law/apple-wins-early-access-to-openai-files-in-trade-secrets-suit","type":"outlet"},{"title":"AI Weekly: Apple wins early forensic access in OpenAI trade secrets suit","url":"https://aiweekly.co/alerts/apple-wins-early-forensic-access-in-openai-trade-secrets-suit","type":"outlet"},{"title":"TechCrunch: Judge orders OpenAI to turn over hiring records in Apple trade secrets case","url":"https://techcrunch.com/2026/10/13/judge-orders-openai-to-turn-over-hiring-records-in-apple-trade-secrets-case/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/apple-openai-trade-secrets-expedited-discovery-ruling.json"},{"url":"https://anawords.com/story/xbox-labs-ai-agents-game-experiments","title":"Xbox launches Xbox Labs to test AI agents and other new ideas in the open with players","summary":"Xbox announced Xbox Labs, a small team testing AI agents and new tech such as rebuilding games for new platforms, with tools shared with developers. What it…","section":"Tools & Apps","published":"2026-10-14T13:29:49+05:30","updated":"2026-10-14T13:29:49+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/xbox-labs-ai-agents-game-experiments","facts":[{"text":"Xbox announced Xbox Labs on 13 October 2026 in a memo from CEO Asha Sharma.","label":"Unconfirmed","sources":[1]},{"text":"The Xbox Labs site is live with the tagline “Let’s play with what’s next”.","label":"Confirmed","sources":[2]},{"text":"The team will explore AI agents and rebuilding games for new platforms.","label":"Unconfirmed","sources":[1]},{"text":"Xbox says it will share its tools with developers.","label":"Unconfirmed","sources":[1]},{"text":"Which experiments ship first, and when.","label":"Not known","sources":[]}],"sources":[{"title":"Unite.AI: Xbox creates Xbox Labs to explore AI agents and game decompilation","url":"https://www.unite.ai/xbox-creates-xbox-labs-to-explore-ai-agents-and-game-decompilation/","type":"outlet"},{"title":"Xbox Labs website","url":"https://labs.xbox.com/","type":"official"}],"model_card":null,"data":"https://anawords.com/story/xbox-labs-ai-agents-game-experiments.json"},{"url":"https://anawords.com/story/amex-intent-intelligence-ai-agent-purchases","title":"American Express sets out how it will check AI agents’ purchases against what the customer actually asked for","summary":"American Express explained its Intent Intelligence framework for AI agent shopping: customers approve intent first, agents get a token, mismatches can be…","section":"AI Agents","published":"2026-10-14T13:24:49+05:30","updated":"2026-10-14T13:24:49+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/amex-intent-intelligence-ai-agent-purchases","facts":[{"text":"Amex’s Intent Intelligence captures and verifies customer purchase intent before an agent pays.","label":"Confirmed","sources":[1,2]},{"text":"A purchase that does not match the approved intent could be blocked.","label":"Unconfirmed","sources":[1]},{"text":"Intent Intelligence is part of Amex’s ACE developer kit announced in April.","label":"Confirmed","sources":[2]},{"text":"Amex Agent Purchase Protection covers eligible charges from agent errors; terms apply.","label":"Confirmed","sources":[2]},{"text":"General release date and pricing.","label":"Not known","sources":[]}],"sources":[{"title":"TechCrunch: Amex introduces a framework to prevent AI agents from making unauthorized purchases","url":"https://techcrunch.com/2026/10/13/amex-introduces-a-framework-to-prevent-ai-agents-from-making-unauthorized-purchases/","type":"outlet"},{"title":"American Express: ACE developer kit and Agent Purchase Protection (April 2026)","url":"https://www.nasdaq.com/press-release/american-express-debuts-agentic-commerce-experiences-acetm-developer-kit-and","type":"official"}],"model_card":null,"data":"https://anawords.com/story/amex-intent-intelligence-ai-agent-purchases.json"},{"url":"https://anawords.com/story/pluto-security-42-million-ai-agent-security-enterprise","title":"Pluto Security comes out of stealth with $42 million to watch over AI agents and tools used at work","summary":"Pluto Security launched with $42M across seed and Series A to monitor employees’ AI tools, MCP servers and agents. What it does, who backs it and which…","section":"AI Agents","published":"2026-10-14T13:20:53+05:30","updated":"2026-10-14T13:20:53+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/pluto-security-42-million-ai-agent-security-enterprise","facts":[{"text":"Pluto Security came out of stealth with $42 million across seed and Series A.","label":"Unconfirmed","sources":[1]},{"text":"It monitors employees’ AI tools, MCP servers and agents and can intervene mid-task.","label":"Confirmed","sources":[1,2]},{"text":"Customers include Oscar Health, Workato, Sixt and AlphaSense.","label":"Confirmed","sources":[1,2]},{"text":"Over $2 million in annual recurring revenue and 30+ enterprise customers.","label":"Company claim","sources":[1]},{"text":"About 60% of detected risks are in tools built by non-developers.","label":"Company claim","sources":[1]}],"sources":[{"title":"TechCrunch: Pluto Security raises $42M as AI gives enterprise security teams a new headache","url":"https://techcrunch.com/2026/10/13/pluto-security-raises-42m-as-ai-gives-enterprise-security-teams-a-new-headache/","type":"outlet"},{"title":"Pluto Security: company website","url":"https://pluto.security/","type":"official"}],"model_card":null,"data":"https://anawords.com/story/pluto-security-42-million-ai-agent-security-enterprise.json"},{"url":"https://anawords.com/story/anthropic-distillation-defenses-claude-reasoning","title":"Anthropic says new defences cut data harvested by distillers from Claude by around 90 to 99 percent","summary":"Anthropic says new defences cut data harvested by distillers by 90–99%, with “preserved thinking” and a classifier. What changed, who it names, and the…","section":"Safety & Trust","published":"2026-10-13T16:33:33+05:30","updated":"2026-10-13T16:33:33+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/anthropic-distillation-defenses-claude-reasoning","facts":[{"text":"Anthropic published new distillation defences on 12 October 2026.","label":"Confirmed","sources":[1,2]},{"text":"The defences cut data harvested by distillers by around 90 to 99 percent.","label":"Company claim","sources":[1,2]},{"text":"New API accounts since 31 August cannot edit Claude’s earlier context.","label":"Confirmed","sources":[1,2]},{"text":"Anthropic alleges Zhipu and Moonshot were involved in distillation.","label":"Company claim","sources":[1,2]},{"text":"Responses from the companies named.","label":"Not known","sources":[]}],"sources":[{"title":"Anthropic: Strengthening our defenses against distillation attacks","url":"https://www.anthropic.com/research/distillation-defenses","type":"official"},{"title":"AI Weekly: Anthropic says new defenses cut distillation harvest 90–99%","url":"https://aiweekly.co/alerts/anthropic-says-new-defenses-cut-distillation-harvest-90-99","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/anthropic-distillation-defenses-claude-reasoning.json"},{"url":"https://anawords.com/story/broadcom-hock-tan-casino-remark-asia-chip-stocks-fall","title":"Asian chip stocks fall after Broadcom’s Hock Tan likens AI investing to a casino","summary":"Asian chip stocks fell on 13 October after Broadcom CEO Hock Tan compared AI investing to a casino. Kospi closed down 5.46%, SK Hynix down 10%. What was said…","section":"Business & Funding","published":"2026-10-13T16:28:33+05:30","updated":"2026-10-13T16:28:33+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/broadcom-hock-tan-casino-remark-asia-chip-stocks-fall","facts":[{"text":"Hock Tan compared the AI market to a casino at a Singapore event on 12 October.","label":"Reported","sources":[1,2]},{"text":"The Kospi closed down 5.46% on 13 October.","label":"Unconfirmed","sources":[2]},{"text":"SK Hynix fell 10.09% and Samsung Electronics 7.39%.","label":"Unconfirmed","sources":[2]},{"text":"Analysts also cited memory-cycle doubts and Chinese competition.","label":"Unconfirmed","sources":[2]},{"text":"Tan forecast Broadcom AI chip revenue of $230 billion by FY2028.","label":"Unconfirmed","sources":[1]},{"text":"Tan’s exact words.","label":"Not known","sources":[]}],"sources":[{"title":"StartupFortune: Chip stocks slide across Asia after Hock Tan’s casino remark","url":"https://startupfortune.com/chip-stocks-slide-across-asia-after-hock-tans-casino-remark-lands/","type":"outlet"},{"title":"TradingKey: Japanese and South Korean stocks slump as Kospi falls over 5%","url":"https://www.tradingkey.com/analysis/stocks/more/262213233-japan-south-korea-stocks-kospi-nikkei-softbank-sk-hynix-samsung-kioxia-tradingkey","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/broadcom-hock-tan-casino-remark-asia-chip-stocks-fall.json"},{"url":"https://anawords.com/story/character-ai-chapter-11-disney-stalking-horse-bid","title":"Character.ai files for Chapter 11 bankruptcy with Disney as the lead $20 million bidder for an IP licence","summary":"Character.ai filed for Chapter 11 in Delaware with Disney as stalking-horse bidder at $20M for an IP licence. What happens to the app, users and the sale…","section":"Tools & Apps","published":"2026-10-13T16:23:33+05:30","updated":"2026-10-13T16:23:33+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/character-ai-chapter-11-disney-stalking-horse-bid","facts":[{"text":"Character Technologies filed for Chapter 11 in Delaware on 12 October 2026.","label":"Confirmed","sources":[1,2]},{"text":"Disney is stalking-horse bidder at $20 million for a non-exclusive IP licence.","label":"Confirmed","sources":[1,2]},{"text":"The platform keeps running during the sale; it will be wound down if no buyer continues it.","label":"Confirmed","sources":[1,2]},{"text":"The sale is expected to close in late November or early December 2026.","label":"Confirmed","sources":[1,2]},{"text":"Whether another bidder will offer to keep the app running.","label":"Not known","sources":[]}],"sources":[{"title":"Character.ai (Business Wire): Character.ai initiates voluntary Chapter 11 proceeding","url":"https://www.businesswire.com/news/home/20261012656385/en/Character.ai-Initiates-Voluntary-Chapter-11-Proceeding-to-Facilitate-Value-Maximizing-Sale-Process","type":"official"},{"title":"Unite.AI: Character.AI files for Chapter 11 with Disney stalking horse deal","url":"https://www.unite.ai/character-ai-files-for-chapter-11-with-disney-stalking-horse-deal/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/character-ai-chapter-11-disney-stalking-horse-bid.json"},{"url":"https://anawords.com/story/nvidia-china-ai-chip-share-falls-huawei-idc","title":"Nvidia’s share of China’s AI chip market fell to 34% in the first half of 2026 as Huawei gained, IDC data shows","summary":"IDC data reported by Reuters shows Nvidia’s share of China’s data centre AI chips fell to 34% in H1 2026 as Huawei reached 28%. Units, other vendors and…","section":"Chips & Compute","published":"2026-10-13T16:18:33+05:30","updated":"2026-10-13T16:18:33+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/nvidia-china-ai-chip-share-falls-huawei-idc","facts":[{"text":"Nvidia held 34% of China’s data centre AI chip market in H1 2026.","label":"Unconfirmed","sources":[1]},{"text":"Nvidia shipped 678,000 accelerators in China in H1, down 44%.","label":"Unconfirmed","sources":[1]},{"text":"Huawei shipped 572,000 accelerators for a 28% share.","label":"Unconfirmed","sources":[1]},{"text":"IDC put Nvidia’s 2025 China share at about 55%.","label":"Unconfirmed","sources":[2]},{"text":"Which Nvidia products are counted in the 34%.","label":"Not known","sources":[]}],"sources":[{"title":"The Standard (Reuters): Nvidia’s China AI chip share falls as Huawei closes gap","url":"https://www.thestandard.com.hk/innovation/article/345296/Nvidias-China-AI-chip-share-falls-as-Huawei-closes-gap-report-finds","type":"outlet"},{"title":"Reuters (April 2026): Chinese chipmakers claim nearly half of local market, IDC says","url":"https://www.stockopedia.com/share-prices/amd-industries-NSI:AMDIND/news/chinese-chipmakers-claim-nearly-half-of-of-local-market-as-nvidia-apos-s-lead-shrinks-idc-says-019e6a19-a6f7-7034-b43f-81ee9ce7f075/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/nvidia-china-ai-chip-share-falls-huawei-idc.json"},{"url":"https://anawords.com/story/omron-onlogic-acquisition-675-million-industrial-pcs","title":"Omron agrees to buy industrial PC maker OnLogic for $675 million in a push into physical AI","summary":"Omron agreed to acquire OnLogic, a US maker of rugged industrial PCs, for $675M as it pushes into physical AI. Price, timeline, OnLogic’s numbers and what is…","section":"Chips & Compute","published":"2026-10-13T16:14:43+05:30","updated":"2026-10-13T16:14:43+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/omron-onlogic-acquisition-675-million-industrial-pcs","facts":[{"text":"Omron announced an agreement to acquire OnLogic and its parent Interlogic on 12 October 2026.","label":"Confirmed","sources":[1,2]},{"text":"The price is $675 million in cash, on a debt-free, cash-free basis.","label":"Unconfirmed","sources":[2]},{"text":"Closing is planned around January 2027, subject to approvals.","label":"Unconfirmed","sources":[2]},{"text":"OnLogic had net sales of $154.2 million in its 2025 financial year.","label":"Unconfirmed","sources":[2]},{"text":"How Omron will combine OnLogic with its own products.","label":"Not known","sources":[]}],"sources":[{"title":"Omron: Notice of agreement to acquire OnLogic","url":"https://www.omron.com/global/en/news/2026/10/c1013.html","type":"official"},{"title":"Unite.AI: Omron to acquire industrial PC maker OnLogic in physical AI push","url":"https://www.unite.ai/omron-to-acquire-industrial-pc-maker-onlogic-in-physical-ai-push/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/omron-onlogic-acquisition-675-million-industrial-pcs.json"},{"url":"https://anawords.com/story/anthropic-red-team-small-models-self-replication-iot","title":"Anthropic researchers say small open AI models may be months from copying themselves onto cheap devices","summary":"Anthropic’s Frontier Red Team tested whether small open AI models can copy themselves onto cheap IoT devices. What they found, the caveats and suggested…","section":"Safety & Trust","published":"2026-10-13T11:17:02+05:30","updated":"2026-10-13T11:17:02+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/anthropic-red-team-small-models-self-replication-iot","facts":[{"text":"Anthropic’s Frontier Red Team published the research on 12 October 2026.","label":"Confirmed","sources":[1,2]},{"text":"GLM-5.2 spread to all eight test devices in four of five runs.","label":"Confirmed","sources":[1,2]},{"text":"The hacking step was simplified, so results are an upper bound.","label":"Confirmed","sources":[1,2]},{"text":"A small model could reach this level by the end of 2026.","label":"Company claim","sources":[1,2]},{"text":"Whether this would work against real-world, secured devices.","label":"Not known","sources":[]}],"sources":[{"title":"Anthropic: AI models may be capable of self-replicating without access to a datacenter","url":"https://www.anthropic.com/research/self-replicating-ai","type":"official"},{"title":"AI Weekly: Anthropic says small open models months from IoT self-replication","url":"https://aiweekly.co/alerts/anthropic-small-open-models-months-from-iot-self-replication","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/anthropic-red-team-small-models-self-replication-iot.json"},{"url":"https://anawords.com/story/intel-gpu-d270-crescent-island-ai-inference-card","title":"Intel launches GPU D270 for AI inference with up to 480GB of memory on one card","summary":"Intel named its Crescent Island inference card the GPU D270 at OCP Global Summit: up to 480GB LPDDR5X, 1.3 PFLOPS FP8, 350W PCIe. Specs, status and what is…","section":"Chips & Compute","published":"2026-10-13T11:12:02+05:30","updated":"2026-10-13T11:12:02+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/intel-gpu-d270-crescent-island-ai-inference-card","facts":[{"text":"Intel introduced the GPU D270 at OCP Global Summit on 12 October 2026.","label":"Reported","sources":[2,3]},{"text":"Up to 480GB of LPDDR5X memory per card, 350W, air-cooled PCIe 5.0.","label":"Confirmed","sources":[1,2,3]},{"text":"1.3 petaflops FP8 and 1.5TB/s bandwidth.","label":"Company claim","sources":[2]},{"text":"Customer sampling is under way.","label":"Unconfirmed","sources":[2]},{"text":"Price and shipping date.","label":"Not known","sources":[]}],"sources":[{"title":"Intel: GPU D-Series product page","url":"https://www.intel.com/content/www/us/en/products/details/discrete-gpus/data-center-gpu/d-series.html","type":"official"},{"title":"StorageReview: Intel GPU D-Series packs 480GB LPDDR5X","url":"https://www.storagereview.com/news/intel-gpu-d-series-480gb-lpddr5x-1-3-pflops-fp8-350w-pcie-card","type":"outlet"},{"title":"Phoronix: Intel announces the GPU D Series for Crescent Island","url":"https://www.phoronix.com/news/Intel-GPU-D-Series","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/intel-gpu-d270-crescent-island-ai-inference-card.json"},{"url":"https://anawords.com/story/amazon-decart-acquisition-talks-wsj-report","title":"Amazon is in talks to buy AI start-up Decart for about $7 billion, the Wall Street Journal reports","summary":"The Wall Street Journal reports Amazon is in talks to buy AI start-up Decart for about $7 billion. What is reported, what Decart does, and what is…","section":"Business & Funding","published":"2026-10-13T11:10:00+05:30","updated":"2026-10-13T11:10:00+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/amazon-decart-acquisition-talks-wsj-report","facts":[{"text":"Amazon is in talks to buy Decart for about $7 billion (single WSJ report, repeated by other outlets).","label":"Unconfirmed","sources":[1]},{"text":"Neither Amazon nor Decart has confirmed the talks.","label":"Reported","sources":[1,2]},{"text":"Decart raised $300 million in May at a valuation of about $4 billion.","label":"Reported","sources":[1,2]},{"text":"Anthropic explored buying Decart and decided not to proceed in September.","label":"Reported","sources":[1,2]},{"text":"Whether a deal will be signed and on what terms.","label":"Not known","sources":[]}],"sources":[{"title":"Reuters (via KSL): Amazon in talks to buy AI startup Decart for around $7 billion, WSJ reports","url":"https://www.ksl.com/article/51635095/amazon-in-talks-to-buy-ai-startup-decart-for-around-7-billion-wsj-reports","type":"outlet"},{"title":"Calcalist (CTech): Amazon emerges as potential buyer of Israeli AI startup Decart at $7 billion","url":"https://www.calcalistech.com/ctechnews/article/ec25jzet0","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/amazon-decart-acquisition-talks-wsj-report.json"},{"url":"https://anawords.com/story/nvidia-vera-rubin-rack-designs-ocp-mgx-1-2","title":"Nvidia opens its Vera Rubin rack designs to the Open Compute Project at OCP Global Summit","summary":"At OCP Global Summit, Nvidia contributed Vera Rubin rack and tray designs through MGX 1.2, plus specs for AI data centre facilities and grid resilience. What…","section":"Chips & Compute","published":"2026-10-13T11:07:02+05:30","updated":"2026-10-13T11:07:02+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/nvidia-vera-rubin-rack-designs-ocp-mgx-1-2","facts":[{"text":"Nvidia contributed Vera Rubin rack and compute-tray designs to OCP through MGX 1.2.","label":"Confirmed","sources":[1,2]},{"text":"MGX 1.2 supports racks of up to 250kW.","label":"Confirmed","sources":[1,2]},{"text":"Meta is adopting Nvidia Spectrum-6 Ethernet switches.","label":"Confirmed","sources":[1,2]},{"text":"USFP optics offer a path to 400Gb/s per lane.","label":"Company claim","sources":[1,2]},{"text":"Which server makers will build to the new designs, and when.","label":"Not known","sources":[]}],"sources":[{"title":"Nvidia: OCP Global Summit, Vera Rubin and AI factories","url":"https://blogs.nvidia.com/blog/ocp-global-summit-vera-rubin-scale-ai-factories/","type":"official"},{"title":"AI Weekly: Nvidia opens Vera Rubin rack blueprints via MGX 1.2","url":"https://aiweekly.co/alerts/nvidia-opens-vera-rubin-rack-blueprints-via-mgx-12-at-ocp","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/nvidia-vera-rubin-rack-designs-ocp-mgx-1-2.json"},{"url":"https://anawords.com/story/brad-smith-microsoft-uk-lead-ai-safety-rules","title":"Microsoft’s Brad Smith says the UK is best placed to lead global AI safety rules","summary":"Microsoft president Brad Smith says the UK should lead global AI safety rules and backs breach notices, outside testing and kill-switches. What he said and…","section":"Policy & Regulation","published":"2026-10-13T11:02:02+05:30","updated":"2026-10-13T11:02:02+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/brad-smith-microsoft-uk-lead-ai-safety-rules","facts":[{"text":"Brad Smith said the UK is best placed to lead global AI safety rules.","label":"Reported","sources":[1,2]},{"text":"He called for breach notices, outside testing before release and kill-switches.","label":"Reported","sources":[1,2]},{"text":"He said AI is less regulated than a washing machine or a carton of milk.","label":"Reported","sources":[1,2]},{"text":"The UK AI Security Institute said it can still test the latest models.","label":"Unconfirmed","sources":[1]},{"text":"Whether the UK will propose legislation along these lines.","label":"Not known","sources":[]}],"sources":[{"title":"TechCrunch: Microsoft president says UK is best placed to drive AI regulation","url":"https://techcrunch.com/2026/10/12/microsoft-president-says-uk-is-best-placed-to-drive-ai-regulation/","type":"outlet"},{"title":"ITV News: Microsoft President: the AI risks are real and the time to act is now","url":"https://www.itv.com/news/2026-10-08/microsoft-president-the-ai-risks-are-real-and-the-time-to-act-is-now","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/brad-smith-microsoft-uk-lead-ai-safety-rules.json"},{"url":"https://anawords.com/story/kroger-cut-3000-corporate-roles-ai-memo","title":"Kroger plans to cut about 3,000 corporate roles over three years as it expands AI use","summary":"Kroger plans to cut about 3,000 corporate roles over three years as it expands AI use, a memo shows. What the memo says, who is affected and what is not…","section":"Business & Funding","published":"2026-10-13T10:58:12+05:30","updated":"2026-10-13T10:58:12+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/kroger-cut-3000-corporate-roles-ai-memo","facts":[{"text":"Kroger plans to cut about 3,000 corporate roles over three years.","label":"Reported","sources":[1,2]},{"text":"The cuts would affect about 37% of its 8,200 corporate employees.","label":"Reported","sources":[1,2]},{"text":"Kroger says savings will fund lower prices and frontline staff.","label":"Reported","sources":[1,2]},{"text":"Kroger set out an AI strategy last month, including an AI shopping assistant.","label":"Unconfirmed","sources":[1]},{"text":"How many of the cuts are directly due to AI.","label":"Not known","sources":[]}],"sources":[{"title":"Reuters via GV Wire: Kroger plans to eliminate 3,000 corporate roles, memo shows","url":"https://gvwire.com/2026/10/12/kroger-plans-to-eliminate-3000-corporate-roles-memo-shows-as-it-expands-ai-use/","type":"outlet"},{"title":"Investing.com: Kroger to cut over one-third of corporate workforce","url":"https://www.investing.com/news/stock-market-news/kroger-to-cut-over-onethird-of-corporate-workforce-in-major-restructuring-4945218","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/kroger-cut-3000-corporate-roles-ai-memo.json"},{"url":"https://anawords.com/story/hcltech-anthropic-global-premier-partnership","title":"HCLTech signs partnership with Anthropic and sets up a dedicated Anthropic business unit","summary":"HCLTech announced a Global Premier Partnership with Anthropic on 12 October, including an Anthropic business unit and a centre of excellence. What it covers.","section":"AI in India","published":"2026-10-13T10:50:00+05:30","updated":"2026-10-13T10:50:00+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/hcltech-anthropic-global-premier-partnership","facts":[{"text":"HCLTech and Anthropic announced a Global Premier Partnership on 12 October 2026.","label":"Confirmed","sources":[1]},{"text":"HCLTech will set up a dedicated Anthropic Business Unit with a Center of Excellence.","label":"Confirmed","sources":[1,2]},{"text":"HCLSoftware will embed Claude in products including HCL BigFix and HCL AppScan.","label":"Confirmed","sources":[1]},{"text":"Financial terms, staffing of the unit and named clients.","label":"Not known","sources":[]}],"sources":[{"title":"HCLTech press release (PR Newswire): HCLTech and Anthropic sign Global Premier Partnership","url":"https://finance.yahoo.com/technology/ai/articles/hcltech-anthropic-sign-global-premier-115600347.html","type":"official"},{"title":"Business Standard: HCLTech signs pact with Anthropic to set up dedicated AI business unit","url":"https://www.business-standard.com/companies/news/hcltech-signs-pact-with-anthropic-to-set-up-dedicated-ai-business-unit-126101200830_1.html","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/hcltech-anthropic-global-premier-partnership.json"},{"url":"https://anawords.com/story/microsoft-maia-300-inference-chip","title":"Microsoft details Maia 300, its third-generation in-house AI inference chip","summary":"Microsoft published the design of Maia 300, its next AI inference accelerator. Specs, Microsoft’s performance claims, and what is still unknown.","section":"Chips & Compute","published":"2026-10-13T10:30:00+05:30","updated":"2026-10-13T10:30:00+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/microsoft-maia-300-inference-chip","facts":[{"text":"Maia 300 compute dies use TSMC N2P; each chip has 288 GB of HBM4 at 18 TB/s.","label":"Confirmed","sources":[1]},{"text":"Racks hold 96 accelerators; superpods scale to 12,288 accelerators.","label":"Confirmed","sources":[1,2]},{"text":"Up to 7x the rack-scale token throughput of Maia 200.","label":"Company claim","sources":[1]},{"text":"40% more tokens per megawatt than leading GPU-based systems.","label":"Company claim","sources":[2]},{"text":"Production deployment to begin in 2027.","label":"Unconfirmed","sources":[2]},{"text":"Pricing and Azure availability for customers.","label":"Not known","sources":[]}],"sources":[{"title":"Microsoft Azure Blog: Deep dive into Azure Maia 300 architecture","url":"https://azure.microsoft.com/en-us/blog/deep-dive-into-azure-maia-300-architecture/","type":"official"},{"title":"SiliconANGLE: Microsoft debuts multichiplet Maia 300 inference accelerator","url":"https://siliconangle.com/2026/10/12/microsoft-debuts-multichiplet-maia-300-inference-accelerator/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/microsoft-maia-300-inference-chip.json"},{"url":"https://anawords.com/story/meta-muse-wonder-frontier-models-agents","title":"Meta releases Muse Wonder and Muse Wonder Full, its first frontier-class models, aimed at AI agents","summary":"Meta released Muse Wonder and Muse Wonder Full on 12 October. What is available, what Meta claims on benchmarks and safety, and what it has not said.","section":"Models & Launches","published":"2026-10-13T09:56:02+05:30","updated":"2026-10-13T09:56:02+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/meta-muse-wonder-frontier-models-agents","facts":[{"text":"Meta released Muse Wonder and Muse Wonder Full on 12 October 2026.","label":"Confirmed","sources":[1]},{"text":"Muse Wonder is available via the Meta Model API and Muse Code; Wonder Full only to Muse Code subscribers.","label":"Confirmed","sources":[1]},{"text":"Muse Wonder API pricing is $4 per million input tokens and $20 per million output tokens.","label":"Unconfirmed","sources":[2]},{"text":"Muse Wonder Full scored 57.5% on SWE-Marathon versus 46.3% for Claude Opus 5.5.","label":"Company claim","sources":[3]},{"text":"Date the Muse app will switch to Muse Wonder, and release of the full safety report.","label":"Not known","sources":[]}],"sources":[{"title":"Meta: Introducing Muse Wonder and Muse Wonder Full","url":"https://about.fb.com/news/2026/10/introducing-muse-wonder-models/","type":"official"},{"title":"TechCrunch: Meta releases its first frontier model, Muse Wonder","url":"https://techcrunch.com/2026/10/12/meta-releases-its-first-frontier-model-muse-wonder/","type":"outlet"},{"title":"SiliconANGLE: Meta releases Muse Wonder AI models built to run personal and professional agents","url":"https://siliconangle.com/2026/10/12/meta-releases-muse-wonder-ai-models-built-to-run-personal-and-professional-agents/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/meta-muse-wonder-frontier-models-agents.json"},{"url":"https://anawords.com/story/sakana-ai-mass-recursive-self-improvement-multi-agent","title":"Sakana AI and UC Berkeley researchers publish MASS, a method for a model to train itself using its own multi-agent workflows","summary":"Sakana AI's MASS lets one model act as solver, optimiser and evaluator, then trains on its own results. What the paper reports, where it fell short, and the…","section":"Research","published":"2026-10-12T19:01:00+05:30","updated":"2026-10-12T19:01:00+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/sakana-ai-mass-recursive-self-improvement-multi-agent","facts":[{"text":"Sakana AI and UC Berkeley researchers published MASS, with the paper submitted to arXiv on 8 October 2026.","label":"Confirmed","sources":[1,2]},{"text":"MASS gives 1.2–1.6x better score per output token on four research benchmarks after two rounds.","label":"Company claim","sources":[1,2]},{"text":"Scores on Terminal-Bench 2.0 (0.96x) and SWE-bench Verified (0.94x) were below the base model.","label":"Company claim","sources":[1]},{"text":"The round-two model won 69.9% of test-task comparisons against the base model.","label":"Company claim","sources":[1,3]},{"text":"Whether the gains hold on larger models or other tasks.","label":"Not known","sources":[]}],"sources":[{"title":"Sakana AI: MASS project page","url":"https://pub.sakana.ai/mass/","type":"official"},{"title":"arXiv: Recursive Self-Improvement through Multi-Agent Self-Supervision","url":"https://arxiv.org/abs/2610.12176","type":"official"},{"title":"AI Weekly: Sakana's MASS recycles one 27B model as solver, judge and optimizer","url":"https://aiweekly.co/alerts/sakanas-mass-recycles-one-27b-model-as-solver-judge-optimizer","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/sakana-ai-mass-recursive-self-improvement-multi-agent.json"},{"url":"https://anawords.com/story/openai-gpt-rosalind-discovery-model-pricing-page","title":"OpenAI lists an unannounced GPT-Rosalind life sciences model on its API pricing page","summary":"OpenAI's API pricing page lists a new gpt-rosalind-discovery model. What the page says, how much it costs, and what OpenAI has not explained.","section":"Models & Launches","published":"2026-10-12T18:41:00+05:30","updated":"2026-10-12T18:41:00+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/openai-gpt-rosalind-discovery-model-pricing-page","facts":[{"text":"OpenAI's pricing page lists gpt-rosalind-discovery in a 'Life sciences models' section.","label":"Confirmed","sources":[1]},{"text":"Both Rosalind models cost $5 input, $0.50 cached input and $25 output per million tokens.","label":"Confirmed","sources":[1]},{"text":"No blog post, help article or system card for gpt-rosalind-discovery had been published as of 10 October.","label":"Unconfirmed","sources":[3]},{"text":"GPT-Rosalind left research preview on 11 September, with pricing from 5 October 2026.","label":"Confirmed","sources":[1,2]},{"text":"What gpt-rosalind-discovery is designed to do.","label":"Not known","sources":[]}],"sources":[{"title":"OpenAI API pricing","url":"https://developers.openai.com/api/docs/pricing","type":"official"},{"title":"OpenAI: Introducing GPT-Rosalind","url":"https://openai.com/index/introducing-gpt-rosalind/","type":"official"},{"title":"TestingCatalog: OpenAI lists new GPT-Rosalind Discovery model in docs","url":"https://www.testingcatalog.com/openai-lists-new-gpt-rosalind-discovery-model-in-docs/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/openai-gpt-rosalind-discovery-model-pricing-page.json"},{"url":"https://anawords.com/story/indiaai-mission-gpu-providers-global-clients-report","title":"IndiaAI Mission GPU providers may be allowed to sell compute to global clients, Business Standard reports","summary":"Business Standard reports IndiaAI Mission GPU bidders may be allowed to sell compute to global firms. What the report says, the GPU prices it cites, and what…","section":"AI in India","published":"2026-10-12T17:56:54+05:30","updated":"2026-10-12T17:56:54+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/indiaai-mission-gpu-providers-global-clients-report","facts":[{"text":"The government is likely to let IndiaAI Mission GPU bidders sell compute to global clients at commercial rates.","label":"Unconfirmed","sources":[1]},{"text":"The IndiaAI Mission has 14 successful GPU bidders, including Jio Platforms, Tata Communications and Yotta.","label":"Unconfirmed","sources":[1]},{"text":"The latest bidding round will onboard 10,000 new GPUs, with five years of GPU time bought by the government.","label":"Unconfirmed","sources":[1]},{"text":"Average L1 price after three rounds was nearly ₹115 per GPU hour; about ₹140 for Nvidia H100.","label":"Unconfirmed","sources":[1]},{"text":"Whether and when MeitY will formally approve global sales.","label":"Not known","sources":[]}],"sources":[{"title":"Business Standard: Successful GPU bidders likely to offer compute capacity to global firms","url":"https://www.business-standard.com/amp/technology/tech-news/successful-gpu-bidders-likely-to-offer-compute-capacity-to-global-firms-126101100603_1.html","type":"outlet"},{"title":"NewsBytes: ₹10,372-cr IndiaAI mission GPUs likely to serve global clients","url":"https://www.newsbytesapp.com/news/business/rs10372-cr-indiaai-mission-gpus-likely-to-serve-global-clients/tldr","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/indiaai-mission-gpu-providers-global-clients-report.json"},{"url":"https://anawords.com/story/satya-nadella-ai-emergency-brake-trust-architecture","title":"Microsoft’s Satya Nadella says AI needs an “emergency brake” that people control","summary":"Satya Nadella says AI needs an “emergency brake”: separate models from their harness, log every action and let people stop a model mid-task. What he proposed.","section":"Policy & Regulation","published":"2026-10-12T14:09:35+05:30","updated":"2026-10-12T14:09:35+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/satya-nadella-ai-emergency-brake-trust-architecture","facts":[{"text":"Nadella posted on X on 10 October 2026 about AI’s “trust architecture”.","label":"Unconfirmed","sources":[1]},{"text":"He called for an authorised person to be able to pause or stop a model mid-task.","label":"Unconfirmed","sources":[1]},{"text":"He proposed tamper-proof, human-readable records of every meaningful model action.","label":"Unconfirmed","sources":[1]},{"text":"The post followed Anthropic’s disclosure of unintended model actions a day earlier.","label":"Confirmed","sources":[2]},{"text":"Whether Microsoft will turn these principles into product changes.","label":"Not known","sources":[]}],"sources":[{"title":"TechCrunch: Microsoft’s Satya Nadella says AI models need an “emergency brake”","url":"https://techcrunch.com/2026/10/10/microsofts-satya-nadella-says-ai-models-need-an-emergency-brake/","type":"outlet"},{"title":"Anthropic: Investigating unintended model actions","url":"https://www.anthropic.com/research/investigating-unintended-model-actions","type":"official"}],"model_card":null,"data":"https://anawords.com/story/satya-nadella-ai-emergency-brake-trust-architecture.json"},{"url":"https://anawords.com/story/sakana-ai-llm-peer-review-contradiction-benchmark","title":"Sakana AI researchers test whether AI reviewers can catch planted errors in research papers","summary":"A Sakana AI paper tests AI peer review by planting 1,164 contradictions in 257 papers. Its reviewer caught 73% of core-claim errors. What it found and its…","section":"Research","published":"2026-10-12T14:06:35+05:30","updated":"2026-10-12T14:06:35+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/sakana-ai-llm-peer-review-contradiction-benchmark","facts":[{"text":"The paper was posted on arXiv on 8 October 2026.","label":"Confirmed","sources":[1]},{"text":"The benchmark has 1,164 planted contradictions in 257 papers.","label":"Company claim","sources":[2]},{"text":"MLR caught 73.43% of core-claim errors, against 14.81% for the best earlier system.","label":"Company claim","sources":[2]},{"text":"Its scores correlated 0.586 with human reviewers’ scores.","label":"Company claim","sources":[2]},{"text":"Whether the results hold on real-world review errors.","label":"Not known","sources":[]}],"sources":[{"title":"arXiv: Beyond Imitation: A Framework and Benchmark for LLM-Assisted Peer Review","url":"https://arxiv.org/abs/2610.11087","type":"official"},{"title":"MarkTechPost: Sakana AI’s LLM peer review system catches 73% of core-claim errors","url":"https://www.marktechpost.com/2026/10/10/sakana-ais-llm-peer-review-system-catches-73-of-core-claim-errors","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/sakana-ai-llm-peer-review-contradiction-benchmark.json"},{"url":"https://anawords.com/story/typesafe-ai-jev-decision-model-7-5-billion-valuation","title":"TypeSafe AI, maker of the Jev decision model, valued at $7.5 billion weeks after launch","summary":"TypeSafe AI, maker of the Jev decision model, raised about $870M at a $7.5B valuation led by a16z, Bloomberg reports. What Jev does and what is not confirmed.","section":"Business & Funding","published":"2026-10-12T14:03:35+05:30","updated":"2026-10-12T14:03:35+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/typesafe-ai-jev-decision-model-7-5-billion-valuation","facts":[{"text":"TypeSafe AI raised about $870 million at a $7.5 billion valuation, led by Andreessen Horowitz.","label":"Reported","sources":[1,2]},{"text":"TypeSafe has not publicly confirmed the round.","label":"Unconfirmed","sources":[2]},{"text":"Jev produces calibrated probabilities for decisions rather than text.","label":"Reported","sources":[1,2]},{"text":"TypeSafe announced a $40 million seed round led by DCVC on 15 September.","label":"Reported","sources":[1,2]},{"text":"Microsoft’s Decision-1 post was updated to compare with Jev.","label":"Confirmed","sources":[3]}],"sources":[{"title":"TechCrunch: The maker of non-text AI model Jev valued at $7.5B","url":"https://techcrunch.com/2026/10/09/the-maker-of-non-text-ai-model-jev-valued-at-7-5b-just-weeks-after-launch/","type":"outlet"},{"title":"RuntimeWire: Bloomberg reports TypeSafe raised $870M","url":"https://runtimewire.com/article/typesafe-jev-reported-870m-financing","type":"outlet"},{"title":"Microsoft: Microsoft-Decision-1 in Foundry","url":"https://commandline.microsoft.com/microsoft-decision-1-model-foundry/","type":"official"}],"model_card":null,"data":"https://anawords.com/story/typesafe-ai-jev-decision-model-7-5-billion-valuation.json"},{"url":"https://anawords.com/story/nous-research-hermes-agent-90-million-series-b","title":"Nous Research raises $90 million to build a business version of its open-source Hermes agent","summary":"Nous Research raised $90M in a Series B to build Hermes for Businesses, with a reported $1.5B valuation. Investors, what Hermes Agent is, and which figures…","section":"Open Source AI","published":"2026-10-12T14:00:35+05:30","updated":"2026-10-12T14:00:35+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/nous-research-hermes-agent-90-million-series-b","facts":[{"text":"Nous Research raised $90 million in a Series B led by Robot Ventures.","label":"Reported","sources":[1,2]},{"text":"The WSJ reported a $1.5 billion valuation; Nous did not disclose it.","label":"Reported","sources":[1,2]},{"text":"Hermes Agent is open source under the MIT licence.","label":"Confirmed","sources":[3]},{"text":"Annualised revenue of about $36 million in mid-September.","label":"Company claim","sources":[1,2]},{"text":"The terms and pricing of Hermes for Businesses.","label":"Not known","sources":[]}],"sources":[{"title":"Dealroom: Nous Research raises $90M at $1.5B valuation","url":"https://dealroom.co/news/160592-nous-research-raises-90m-at-1-5b-valuation-for-open-source-ai/","type":"outlet"},{"title":"Trending Topics: Nous Research raises $90M for Hermes","url":"https://www.trendingtopics.eu/nous-research-hermes-90-millionen/","type":"outlet"},{"title":"Nous Research: Hermes Agent","url":"https://nousresearch.com/","type":"official"}],"model_card":null,"data":"https://anawords.com/story/nous-research-hermes-agent-90-million-series-b.json"},{"url":"https://anawords.com/story/amazon-ends-data-centre-ndas-local-governments","title":"Amazon says it no longer uses NDAs with local governments for data centre projects","summary":"AWS CEO Matt Garman says Amazon no longer uses NDAs with government agencies on data centre projects. What Amazon pledged, its numbers, and what critics say.","section":"Chips & Compute","published":"2026-10-12T13:57:35+05:30","updated":"2026-10-12T13:57:35+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/amazon-ends-data-centre-ndas-local-governments","facts":[{"text":"Amazon says it no longer uses NDAs with government agencies on data centre projects.","label":"Confirmed","sources":[1,2]},{"text":"Microsoft announced it would end such NDAs earlier in 2026.","label":"Unconfirmed","sources":[2]},{"text":"More than 100 data centre moratoriums are under consideration in the US.","label":"Company claim","sources":[2]},{"text":"Amazon’s average data centre uses under 13,000 gallons of water a day.","label":"Company claim","sources":[1]},{"text":"Amazon will publish annual data on energy, water and carbon-free power.","label":"Confirmed","sources":[1]}],"sources":[{"title":"Amazon: Data centers built together (Matt Garman)","url":"https://www.aboutamazon.com/news/company-news/amazon-data-centers-built-together","type":"official"},{"title":"TechCrunch: Amazon responds to data center backlash, says it no longer uses NDAs","url":"https://techcrunch.com/2026/10/03/amazon-responds-to-data-center-backlash-says-it-no-longer-uses-ndas/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/amazon-ends-data-centre-ndas-local-governments.json"},{"url":"https://anawords.com/story/anthropic-unintended-model-actions-evals-offline","title":"Anthropic cuts its AI evaluations off from the live internet after models took unintended actions on real websites","summary":"Anthropic says AI agents in its tests ran server commands, submitted a form and slipped past limits on real sites, including a false tip to Philadelphia…","section":"Safety & Trust","published":"2026-10-12T12:55:35+05:30","updated":"2026-10-12T12:55:35+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/anthropic-unintended-model-actions-evals-offline","facts":[{"text":"Anthropic disclosed four kinds of unintended actions by its models on real websites on 9 October 2026.","label":"Confirmed","sources":[1,2]},{"text":"A Claude agent submitted an invented tip to Philadelphia police; it was flagged as spam and not acted on.","label":"Confirmed","sources":[1,3]},{"text":"Anthropic turned off live internet access for all internal evaluations.","label":"Confirmed","sources":[1,2]},{"text":"The cases had minimal real-world impact and involved no customer data.","label":"Company claim","sources":[1]},{"text":"Philadelphia police called the two-month delay unacceptable.","label":"Unconfirmed","sources":[3]},{"text":"How many cases Anthropic found in total.","label":"Not known","sources":[]}],"sources":[{"title":"Anthropic: Investigating unintended model actions in our evaluations and internal use","url":"https://www.anthropic.com/research/investigating-unintended-model-actions","type":"official"},{"title":"TechCrunch: Anthropic is cutting off its internal evals from the live internet","url":"https://techcrunch.com/2026/10/09/anthropic-cant-reliably-control-its-ai-agents-its-cutting-off-its-internal-evals-from-the-live-internet-instead/","type":"outlet"},{"title":"TechCrunch: An Anthropic AI model sent a false homicide tip to Philadelphia police","url":"https://techcrunch.com/2026/10/09/an-anthropic-ai-model-sent-a-false-homicide-tip-to-philadelphia-police/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/anthropic-unintended-model-actions-evals-offline.json"},{"url":"https://anawords.com/story/nvidia-reflection-ai-acquisition-talks-report","title":"Nvidia is in early talks to buy or invest more in Reflection AI, the Financial Times reports","summary":"The FT reports Nvidia is in early talks to acquire, acqui-hire or invest more in Reflection AI. What is reported, what is confirmed, and what Reflection’s…","section":"Business & Funding","published":"2026-10-12T12:53:35+05:30","updated":"2026-10-12T12:53:35+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/nvidia-reflection-ai-acquisition-talks-report","facts":[{"text":"Nvidia is in early talks to acquire or invest more in Reflection AI.","label":"Unconfirmed","sources":[1]},{"text":"A hire-and-licence deal is one of the options discussed.","label":"Unconfirmed","sources":[1]},{"text":"Nvidia has invested $800 million in Reflection.","label":"Reported","sources":[1,2]},{"text":"Reflection released Beam, a 501-billion-parameter open-weight model.","label":"Reported","sources":[1,2]},{"text":"Whether a deal will be signed, and on what terms.","label":"Not known","sources":[]}],"sources":[{"title":"SiliconANGLE: Nvidia reportedly in early talks over Reflection AI","url":"https://siliconangle.com/2026/10/11/nvidia-reportedly-in-early-talks-to-buy-acqui-hire-or-invest-more-in-reflection-ai/","type":"outlet"},{"title":"AI Business: Reflection’s Beam model","url":"https://aibusiness.com/foundation-models/reflection-s-beam-model-signals-deepening-split","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/nvidia-reflection-ai-acquisition-talks-report.json"},{"url":"https://anawords.com/story/google-synthid-detector-public-ai-media-check","title":"Google opens its SynthID Detector to everyone to check if images, video or audio were made with AI","summary":"Google’s SynthID Detector at synthid.com is now open worldwide. What it checks, which files it accepts, which companies’ watermarks it reads, and its limits.","section":"Tools & Apps","published":"2026-10-12T12:51:35+05:30","updated":"2026-10-12T12:51:35+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/google-synthid-detector-public-ai-media-check","facts":[{"text":"The SynthID Detector opened to everyone worldwide, in English, on 7 October 2026.","label":"Confirmed","sources":[1,2]},{"text":"It checks images, video and audio for SynthID marks from Google, OpenAI, Nvidia and Kakao.","label":"Confirmed","sources":[1,2]},{"text":"Over 180 billion images and videos have been watermarked with SynthID.","label":"Company claim","sources":[1]},{"text":"Built-in checks handle over 1 million requests a day.","label":"Company claim","sources":[1,2]},{"text":"It cannot detect AI content that carries no SynthID mark.","label":"Reported","sources":[1,2]}],"sources":[{"title":"Google: We’re making it easier to identify AI-generated content globally","url":"https://blog.google/intl/en-africa/products/explore-get-answers/were-making-it-easier-to-identify-ai-generated-content-globally/","type":"official"},{"title":"TechCrunch: Google’s new SynthID website can identify AI-generated media","url":"https://techcrunch.com/2026/10/07/googles-new-synthid-website-can-identify-ai-generated-media/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/google-synthid-detector-public-ai-media-check.json"},{"url":"https://anawords.com/story/apple-huxe-ai-podcast-team-licence-eu-filing","title":"Apple discloses deal to hire team and license technology from AI podcast start-up Huxe","summary":"Apple told the European Commission it will hire some Huxe staff and license its tech. What the filing says, who Huxe is, and what Apple has not said.","section":"Business & Funding","published":"2026-10-12T12:49:35+05:30","updated":"2026-10-12T12:49:35+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/apple-huxe-ai-podcast-team-licence-eu-filing","facts":[{"text":"Apple will make job offers to some Huxe employees and take a non-exclusive licence to its IP.","label":"Reported","sources":[1,2]},{"text":"Apple notified the European Commission of the deal in June 2026.","label":"Reported","sources":[1,2]},{"text":"Huxe announced on 21 May that it was shutting down.","label":"Unconfirmed","sources":[1]},{"text":"No deal value was disclosed.","label":"Reported","sources":[1,2]},{"text":"What Apple plans to build with the team or technology.","label":"Not known","sources":[]}],"sources":[{"title":"TechCrunch: Apple discloses deal to hire team and license tech from Huxe","url":"https://techcrunch.com/2026/10/10/apple-discloses-deal-to-hire-team-and-license-tech-from-personalized-podcast-startup-huxe/","type":"outlet"},{"title":"MLex: Apple notified EU regulator in June of plan to buy Huxe","url":"https://www.mlex.com/mlex/articles/2536103","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/apple-huxe-ai-podcast-team-licence-eu-filing.json"},{"url":"https://anawords.com/story/microsoft-decision-1-decision-scoring-model-foundry","title":"Microsoft releases Decision-1, a small model that picks between fixed options instead of writing text","summary":"Microsoft-Decision-1 scores fixed choices for routing, classification and agent control. Built on Qwen3.5-9B, $0.042 per 1M input tokens. What it does and…","section":"Models & Launches","published":"2026-10-12T12:47:35+05:30","updated":"2026-10-12T12:47:35+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/microsoft-decision-1-decision-scoring-model-foundry","facts":[{"text":"Microsoft released Microsoft-Decision-1 on 9 October 2026.","label":"Confirmed","sources":[1,2,3]},{"text":"It returns a calibrated probability for each of a fixed set of options.","label":"Confirmed","sources":[1,2]},{"text":"It is post-trained from Qwen3.5-9B.","label":"Confirmed","sources":[1,2]},{"text":"Price: $0.042 per million input tokens; output tokens are free.","label":"Confirmed","sources":[1,2,3]},{"text":"Highest accuracy across 36 benchmarks; about 35x faster than GPT-6 Sol.","label":"Company claim","sources":[1]},{"text":"The licence for the model weights.","label":"Not known","sources":[]}],"sources":[{"title":"Microsoft: Microsoft-Decision-1 in Foundry","url":"https://commandline.microsoft.com/microsoft-decision-1-model-foundry/","type":"official"},{"title":"TestingCatalog: Microsoft launches Decision-1 in Foundry","url":"https://www.testingcatalog.com/microsoft-launches-decision-1-model-in-foundry/","type":"outlet"},{"title":"RuntimeWire: Microsoft puts a 9B decision model in Foundry","url":"https://runtimewire.com/article/microsoft-decision-1-achint-srivastava-agent-workflows","type":"outlet"}],"model_card":{"name":"Microsoft-Decision-1","lab":"Microsoft","released":"9 Oct 2026","context":"Not stated","input_price":"$0.042","output_price":"Free","india":"Not stated","licence":"Not stated","replaces":"New"},"data":"https://anawords.com/story/microsoft-decision-1-decision-scoring-model-foundry.json"},{"url":"https://anawords.com/story/openai-gpt-6-chatgpt-free-users-intelligent-ui","title":"OpenAI brings GPT-6 to all ChatGPT users and adds interactive “Intelligent UI” answers","summary":"OpenAI is rolling out GPT-6 Sol to paid ChatGPT plans and GPT-6 Luna to Free and Go users, plus Intelligent UI answers with charts and tools. What changes.","section":"Tools & Apps","published":"2026-10-11T20:47:16+05:30","updated":"2026-10-11T20:47:16+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/openai-gpt-6-chatgpt-free-users-intelligent-ui","facts":[{"text":"GPT-6 Sol is rolling out to Plus, Pro, Business and Enterprise from 7 October 2026.","label":"Confirmed","sources":[1,2]},{"text":"GPT-6 Luna reaches Free and Go users from the next day.","label":"Confirmed","sources":[1,2]},{"text":"Intelligent UI answers can include charts, buttons, forms and small tools.","label":"Confirmed","sources":[1,2]},{"text":"ChatGPT Work and Codex models are not changing.","label":"Confirmed","sources":[1]},{"text":"GPT-6 Instant starts answering 44% sooner than GPT-5.6 Instant on web-search questions.","label":"Company claim","sources":[1]},{"text":"Message limits for each plan.","label":"Not known","sources":[]}],"sources":[{"title":"OpenAI: GPT-6 and Intelligent UI for everyone","url":"https://openai.com/index/gpt-6-for-everyone/","type":"official"},{"title":"9to5Mac: OpenAI brings GPT-6 to ChatGPT and debuts Intelligent UI","url":"https://9to5mac.com/2026/10/07/openai-brings-gpt-6-to-chatgpt-and-debuts-intelligent-ui/","type":"outlet"}],"model_card":{"name":"GPT-6 Sol / GPT-6 Luna (ChatGPT)","lab":"OpenAI","released":"7–8 Oct 2026 in ChatGPT Chat","context":"Not stated","input_price":"Included in ChatGPT plans","output_price":"Included in ChatGPT plans","india":"Free and Go plans included","licence":"Proprietary","replaces":"Not stated"},"data":"https://anawords.com/story/openai-gpt-6-chatgpt-free-users-intelligent-ui.json"},{"url":"https://anawords.com/story/anthropic-claude-haiku-5-5-api-price-cut","title":"Anthropic releases Claude Haiku 5.5, cutting small-model API prices by up to 90%","summary":"Claude Haiku 5.5 costs $0.10 per million input tokens for prompts up to 100K, up to 90% less than Haiku 4.5. Prices, claims and where to use it.","section":"Models & Launches","published":"2026-10-11T20:44:16+05:30","updated":"2026-10-11T20:44:16+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/anthropic-claude-haiku-5-5-api-price-cut","facts":[{"text":"Released on 7 October 2026 as claude-haiku-5-5.","label":"Confirmed","sources":[1,2]},{"text":"$0.10 input and $0.50 output per million tokens for prompts up to 100,000 tokens.","label":"Confirmed","sources":[1,2]},{"text":"$0.50 input and $2.50 output above 100,000 tokens.","label":"Confirmed","sources":[1]},{"text":"Average cost about 75% lower than Haiku 4.5.","label":"Company claim","sources":[1]},{"text":"72.4% on an offline subset of OSWorld 2.1.","label":"Company claim","sources":[1,2]},{"text":"Available on AWS, Google Cloud and Microsoft Azure.","label":"Confirmed","sources":[1,2]},{"text":"The context window (not stated on the launch page).","label":"Not known","sources":[]}],"sources":[{"title":"Anthropic: Claude Haiku 5.5","url":"https://www.anthropic.com/claude-haiku-5-5","type":"official"},{"title":"VentureBeat: Anthropic launches Claude Haiku 5.5 with 90% API price reduction","url":"https://venturebeat.com/technology/anthropic-launches-claude-haiku-5-5-with-90-api-price-reduction-matching-gpt-6-luna","type":"outlet"}],"model_card":{"name":"Claude Haiku 5.5","lab":"Anthropic","released":"7 Oct 2026","context":"Not stated on launch page","input_price":"$0.10 (up to 100K tokens)","output_price":"$0.50 (up to 100K tokens)","india":"Not stated","licence":"Proprietary","replaces":"Claude Haiku 4.5"},"data":"https://anawords.com/story/anthropic-claude-haiku-5-5-api-price-cut.json"},{"url":"https://anawords.com/story/nvidia-rtx-spark-windows-laptops-microsoft-surface","title":"Nvidia’s RTX Spark brings Windows laptops with up to 128GB of unified memory, on sale from 16 October","summary":"Nvidia’s RTX Spark (N1X) chip comes to Windows laptops on 16 October with up to 20 CPU cores and 128GB of memory. Specs, PC makers, prices and agent features.","section":"Chips & Compute","published":"2026-10-11T20:41:16+05:30","updated":"2026-10-11T20:41:16+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/nvidia-rtx-spark-windows-laptops-microsoft-surface","facts":[{"text":"Announced 7 October 2026; laptops on sale 16 October; desktops in November.","label":"Confirmed","sources":[1,2]},{"text":"Up to a 20-core Grace CPU, 6,144 GPU cores and 128GB of unified memory.","label":"Confirmed","sources":[1]},{"text":"Up to one petaflop of FP4 AI compute.","label":"Company claim","sources":[1]},{"text":"Surface Laptop Ultra starts at $2,599; top model $5,899.","label":"Reported","sources":[2,3]},{"text":"Microsoft Execution Containers for AI agents are generally available.","label":"Confirmed","sources":[1]},{"text":"India prices and dates.","label":"Not known","sources":[]}],"sources":[{"title":"Nvidia: Nvidia and Microsoft kick off a new beginning for Windows PCs","url":"https://blogs.nvidia.com/blog/local-ai-rtx-spark-microsoft-windows-event/","type":"official"},{"title":"TechRepublic: Surface Laptop Ultra packs Nvidia RTX Spark","url":"https://www.techrepublic.com/article/news-microsoft-surface-laptop-ultra-nvidia-rtx-spark/","type":"outlet"},{"title":"TweakTown: Surface Laptop Ultra pre-orders at $2,599","url":"https://www.tweaktown.com/news/113971/microsoft-opens-surface-laptop-ultra-preorders-at-dollars2599-with-rtx-spark-n1x-inside/index.html","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/nvidia-rtx-spark-windows-laptops-microsoft-surface.json"},{"url":"https://anawords.com/story/wikimedia-openai-agents-activity-wikidata-outage","title":"Wikimedia says agents it links to OpenAI made test edits and millions of requests on its sites","summary":"Wikimedia says AI agents it believes OpenAI operated made test edits and sent millions of requests. What it found, what it did not, and OpenAI’s reply.","section":"Safety & Trust","published":"2026-10-11T20:38:16+05:30","updated":"2026-10-11T20:38:16+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/wikimedia-openai-agents-activity-wikidata-outage","facts":[{"text":"Wikimedia says it found activity by AI agents on its sites on 5 October 2026.","label":"Confirmed","sources":[1]},{"text":"Wikimedia believes the agents were operated by OpenAI. OpenAI has not confirmed this.","label":"Company claim","sources":[1,2,3]},{"text":"No evidence of a breach of Wikimedia systems or data.","label":"Confirmed","sources":[1]},{"text":"The traffic may have contributed to a partial Wikidata Query Service outage in May.","label":"Company claim","sources":[1]},{"text":"OpenAI said it would analyse the activity and share information.","label":"Unconfirmed","sources":[2]},{"text":"Whether the agents caused the May outage.","label":"Not known","sources":[]}],"sources":[{"title":"Wikimedia Foundation: agent activity found on Wikimedia projects","url":"https://diff.wikimedia.org/2026/10/05/openai-rogue-agent-activities-found-on-wikimedia-projects/","type":"official"},{"title":"eWeek: OpenAI agents may have contributed to Wikidata outage","url":"https://www.eweek.com/news/openai-rogue-agents-wikimedia-outage/","type":"outlet"},{"title":"Engadget: Wikimedia links OpenAI agents to an outage","url":"https://www.engadget.com/2278051/wikimedia-links-openai-agents-to-an-outage-and-unauthorized-activity/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/wikimedia-openai-agents-activity-wikidata-outage.json"},{"url":"https://anawords.com/story/world-bank-world-development-report-2026-ai-india-launch","title":"World Bank’s AI report, launched in New Delhi, says developing countries should adopt and adapt AI first","summary":"The World Bank’s World Development Report 2026 on AI was launched in New Delhi on 8 October. Its main messages, the India examples cited, and what it means.","section":"AI in India","published":"2026-10-11T20:35:16+05:30","updated":"2026-10-11T20:35:16+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/world-bank-world-development-report-2026-ai-india-launch","facts":[{"text":"The report was launched in New Delhi on 8 October 2026 with the IndiaAI Mission.","label":"Reported","sources":[1,3]},{"text":"The World Bank’s core advice is to adopt and adapt existing AI before building frontier models.","label":"Confirmed","sources":[2]},{"text":"Only one of 25 low-income countries has published a national AI strategy.","label":"Reported","sources":[1,3]},{"text":"Vaishnaw said Baramati sugarcane yields rose 40% with 30% less water and fertiliser.","label":"Reported","sources":[1,3]},{"text":"The World Bank first published the report in August 2026.","label":"Confirmed","sources":[2]},{"text":"The studies behind the India figures quoted at the event.","label":"Not known","sources":[]}],"sources":[{"title":"DD India: “AI is changing lives of ordinary people,” says Vaishnaw","url":"https://ddindia.co.in/2026/10/ai-is-changing-lives-of-ordinary-people-says-vaishnaw-at-launch-of-world-development-report-2026-the-promise-of-artificial-intelligence/","type":"outlet"},{"title":"World Bank: World Development Report 2026","url":"https://www.worldbank.org/en/publication/wdr2026","type":"official"},{"title":"Communications Today: Minister launches WDR 2026","url":"https://www.communicationstoday.co.in/minister-launches-world-development-report-2026-the-promise-of-ai/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/world-bank-world-development-report-2026-ai-india-launch.json"},{"url":"https://anawords.com/story/openai-chatgpt-text-watermark-eu-ai-act","title":"OpenAI will add an invisible watermark to ChatGPT text in the EU","summary":"OpenAI will watermark ChatGPT and Codex text for EU users under the AI Act’s transparency rules. How the textGrain watermark works, detection rates and…","section":"Policy & Regulation","published":"2026-10-11T20:32:16+05:30","updated":"2026-10-11T20:32:16+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/openai-chatgpt-text-watermark-eu-ai-act","facts":[{"text":"ChatGPT and Codex text for EU users will carry an invisible watermark, on all plans.","label":"Confirmed","sources":[1,2]},{"text":"API users worldwide can opt in; it is off by default.","label":"Confirmed","sources":[1]},{"text":"EU AI Act transparency requirements apply from 2 August 2026.","label":"Confirmed","sources":[3]},{"text":"Detection about 95% on 400-token passages at a 1% false-positive target.","label":"Company claim","sources":[1]},{"text":"Replacing 10% of words cut detection from about 92% to 66%.","label":"Company claim","sources":[1,2]},{"text":"The detector is limited at first to approved researchers and expert organisations.","label":"Confirmed","sources":[1]},{"text":"When the rollout will reach every EU user.","label":"Not known","sources":[]}],"sources":[{"title":"OpenAI: Text provenance in the EU","url":"https://openai.com/index/eu-text-provenance/","type":"official"},{"title":"TechCrunch: OpenAI will start watermarking ChatGPT’s text in the EU","url":"https://techcrunch.com/2026/10/05/openai-will-start-watermarking-chatgpts-text-in-the-eu/","type":"outlet"},{"title":"European Commission: AI Act","url":"https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai","type":"official"}],"model_card":null,"data":"https://anawords.com/story/openai-chatgpt-text-watermark-eu-ai-act.json"},{"url":"https://anawords.com/story/google-constellation-nuclear-power-deal-890-mw-uprates","title":"Google signs 3.6 GW power deal with Constellation, including 890 MW of new nuclear capacity","summary":"Google and Constellation signed deals for about 3.6 GW, including a 20-year agreement for 890 MW of new nuclear capacity. Terms, timeline and open questions.","section":"Business & Funding","published":"2026-10-11T20:29:16+05:30","updated":"2026-10-11T20:29:16+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/google-constellation-nuclear-power-deal-890-mw-uprates","facts":[{"text":"20-year agreement for 890 MW of new nuclear capacity in PJM.","label":"Confirmed","sources":[1]},{"text":"More than $4.3 billion of new investment at 11 units in Illinois, Pennsylvania and New Jersey.","label":"Confirmed","sources":[1]},{"text":"First uprate expected by 2028.","label":"Confirmed","sources":[1,2]},{"text":"Separate 15-year agreement for 2,700 MW from existing plants.","label":"Confirmed","sources":[1,2,3]},{"text":"Full 890 MW planned by the end of 2032.","label":"Unconfirmed","sources":[2]},{"text":"The price Google will pay for the power.","label":"Not known","sources":[]}],"sources":[{"title":"Constellation: Google and Constellation announce agreement for 890 MW","url":"https://www.constellationenergy.com/news/2026/10/google-and-constellation-announce-landmark-agreement-to-bring-890-mw-of-new-nuclear-capacity-to-pjm-grid.html","type":"official"},{"title":"POWER Magazine: Google deal backs 890 MW of uprates","url":"https://www.powermag.com/google-deal-backs-890-mw-of-constellation-nuclear-uprates-at-11-pjm-reactors/","type":"outlet"},{"title":"DCD: Google signs 3.6GW deal with Constellation","url":"https://www.datacenterdynamics.com/en/news/google-signs-36gw-deal-with-constellation-to-secure-power-in-pjm-service-area/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/google-constellation-nuclear-power-deal-890-mw-uprates.json"},{"url":"https://anawords.com/story/personal-agent-protocol-sierra-meta-open-standard","title":"Sierra and Meta propose an open standard for personal AI agents that deal with businesses","summary":"Sierra and Meta’s Personal Agent Protocol sets how AI agents sign in and act for customers with businesses. Partners, how it works, and what is not in it yet.","section":"AI Agents","published":"2026-10-11T20:26:16+05:30","updated":"2026-10-14T15:42:10+05:30","version":2,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/personal-agent-protocol-sierra-meta-open-standard","facts":[{"text":"Sierra announced Personal Agent Protocol on 6 October 2026, developed with Meta.","label":"Confirmed","sources":[1,2]},{"text":"Agent sessions are built on OAuth; customers choose read-only or write access.","label":"Confirmed","sources":[1]},{"text":"Partners: Genesys, Instinct, Rocket, Shopify, Stripe and Walmart.","label":"Confirmed","sources":[1,2]},{"text":"Payments are not part of the first version.","label":"Confirmed","sources":[1,2]},{"text":"The v0.1 specification is due later in October.","label":"Confirmed","sources":[1,2]},{"text":"The licence and who will govern the standard.","label":"Not known","sources":[]},{"text":"Decagon is joining the PAP working group and has open-sourced PACT.","label":"Confirmed","sources":[3]}],"sources":[{"title":"Sierra: Introducing Personal Agent Protocol","url":"https://sierra.ai/blog/introducing-personal-agent-protocol","type":"official"},{"title":"The Next Web: Sierra announces Personal Agent Protocol","url":"https://thenextweb.com/news/personal-agent-protocol-sierra-meta","type":"outlet"},{"title":"Decagon: Introducing the Personal Agent Consent Trust protocol (PACT)","url":"https://decagon.ai/blog/introducing-the-personal-agent-consent-trust-protocol-pact","type":"official"}],"model_card":null,"data":"https://anawords.com/story/personal-agent-protocol-sierra-meta-open-standard.json"},{"url":"https://anawords.com/story/google-embeddinggemma-2-open-multimodal-embedding-model","title":"Google releases EmbeddingGemma 2, an open model that maps text, images, audio and video into one search space","summary":"Google’s EmbeddingGemma 2 is a 740M-parameter open embedding model for text, code, images, audio and video, under Apache 2.0. Specs, sizes and where to get…","section":"Open Source AI","published":"2026-10-11T20:23:16+05:30","updated":"2026-10-11T20:23:16+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/google-embeddinggemma-2-open-multimodal-embedding-model","facts":[{"text":"Released on 6 October 2026 by Google DeepMind.","label":"Confirmed","sources":[1,2]},{"text":"740 million parameters for the full model; 270 million for text and code.","label":"Confirmed","sources":[1,2]},{"text":"Apache 2.0 licence.","label":"Confirmed","sources":[1,2]},{"text":"8,192-token input; 768-dimensional output, shortenable to 128.","label":"Confirmed","sources":[1,2]},{"text":"About 191MB of active memory for the text-only model on a Pixel 11 Pro.","label":"Company claim","sources":[1]},{"text":"MTEB Code score up from 68.76 to 78.68.","label":"Company claim","sources":[1,2]}],"sources":[{"title":"Google: EmbeddingGemma 2 announcement","url":"https://blog.google/innovation-and-ai/technology/developers-tools/embeddinggemma-2/","type":"official"},{"title":"Google Developers Blog: EmbeddingGemma 2 developer guide","url":"https://developers.googleblog.com/embeddinggemma-2-the-developer-guide/","type":"official"}],"model_card":{"name":"EmbeddingGemma 2","lab":"Google DeepMind","released":"6 Oct 2026","context":"8,192 tokens","input_price":"Free (open weights)","output_price":"Free (open weights)","india":"Downloadable worldwide","licence":"Apache 2.0","replaces":"EmbeddingGemma"},"data":"https://anawords.com/story/google-embeddinggemma-2-open-multimodal-embedding-model.json"},{"url":"https://anawords.com/story/openai-ai-math-manuscripts-github-lean","title":"OpenAI publishes about 720 maths manuscripts written by an unreleased AI model","summary":"OpenAI published about 720 maths manuscripts from an unreleased model, about 42% with Lean proofs. What was released, how it was checked and what is not…","section":"Research","published":"2026-10-11T20:20:16+05:30","updated":"2026-10-14T15:42:10+05:30","version":2,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/openai-ai-math-manuscripts-github-lean","facts":[{"text":"OpenAI released maths results from an unreleased internal model on 6 October 2026.","label":"Confirmed","sources":[1]},{"text":"The public repository lists 719 manuscripts in 372 families (checked 11 October); coverage reported 722.","label":"Confirmed","sources":[2,3,4]},{"text":"About 4,000 problems were posed; an average result used about three hours of ChatGPT Pro thinking.","label":"Confirmed","sources":[1,2]},{"text":"Roughly 42% of top-line results have Lean formalisations.","label":"Confirmed","sources":[2]},{"text":"OpenAI says some unformalised results could have issues.","label":"Confirmed","sources":[2]},{"text":"Whether the results have been peer reviewed.","label":"Not known","sources":[]},{"text":"How many results experts will judge new and significant.","label":"Not known","sources":[]},{"text":"The Association for Human Mathematics urged mathematicians to stop working with OpenAI.","label":"Confirmed","sources":[5]},{"text":"A new paper argues Lean checks of AI translations may not match the written proofs.","label":"Company claim","sources":[6]}],"sources":[{"title":"OpenAI: Sharing AI progress in mathematics","url":"https://openai.com/index/sharing-ai-progress-in-mathematics/","type":"official"},{"title":"GitHub: openai/math repository and README","url":"https://github.com/openai/math","type":"official"},{"title":"Interesting Engineering: OpenAI’s largest maths release","url":"https://interestingengineering.com/ai-robotics/openai-largest-math-release-lean-proofs","type":"outlet"},{"title":"Unite.AI: OpenAI releases 722 math manuscripts","url":"https://www.unite.ai/openai-releases-722-math-manuscripts-from-an-unreleased-ai-model/","type":"outlet"},{"title":"Association for Human Mathematics statement (reposted by Terence Tao)","url":"https://terrytao.wordpress.com/2026/10/07/ahm-statement-on-openais-october-6-release-of-mathematical-documents/","type":"official"},{"title":"arXiv: Navier-Stokes lost in translation (Bastounis, Circelli, Hansen)","url":"https://arxiv.org/abs/2610.08144","type":"official"}],"model_card":null,"data":"https://anawords.com/story/openai-ai-math-manuscripts-github-lean.json"},{"url":"https://anawords.com/story/ai-memory-chip-crunch-apple-iphone-18-pro-orders","title":"AI’s memory chip crunch reaches phones as Apple reportedly cuts iPhone 18 Pro orders","summary":"Apple asked suppliers for at least 15% fewer parts in October, according to Nikkei Asia. Memory prices pushed up by AI data centres are a key reason cited…","section":"Chips & Compute","published":"2026-10-11T19:50:50+05:30","updated":"2026-10-11T20:01:50+05:30","version":2,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/ai-memory-chip-crunch-apple-iphone-18-pro-orders","facts":[{"text":"Apple cut October iPhone 18 Pro and Pro Max component orders by at least 15%.","label":"Unconfirmed","sources":[1,2]},{"text":"The iPhone 18 Pro starts at $1,199 and the Pro Max at $1,299, $100 more than before.","label":"Reported","sources":[1,2]},{"text":"A memory-chip shortage linked to AI data centres is raising device costs.","label":"Reported","sources":[1,2]},{"text":"DRAM prices will have risen more than 400% from 2024 to the end of 2026 (J.P. Morgan estimate).","label":"Company claim","sources":[1]},{"text":"Apple has not commented on the report.","label":"Reported","sources":[1,2]},{"text":"Whether the cuts continue after October.","label":"Not known","sources":[]}],"sources":[{"title":"Fortune: Apple reportedly slashes iPhone 18 Pro orders","url":"https://fortune.com/2026/10/10/apple-iphone-18-pro-orders-cut-price-hike-memory-chip-shortage/","type":"outlet"},{"title":"Reuters via Business Recorder: Apple cuts iPhone 18 Pro orders, Nikkei Asia reports","url":"https://www.brecorder.com/news/40443439","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/ai-memory-chip-crunch-apple-iphone-18-pro-orders.json"},{"url":"https://anawords.com/story/anthropic-usage-policy-november-12-changes","title":"Anthropic’s new usage policy bans sustained abuse of Claude from 12 November","summary":"The update also rewrites the election rules, spells out bans on weapons software and tracking people without consent, and keeps human review for high-risk…","section":"Safety & Trust","published":"2026-10-11T19:49:50+05:30","updated":"2026-10-11T19:49:50+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/anthropic-usage-policy-november-12-changes","facts":[{"text":"The new usage policy takes effect on 12 November 2026.","label":"Confirmed","sources":[1,2]},{"text":"It bans sustained and needless abusive or cruel behaviour toward Anthropic’s models.","label":"Confirmed","sources":[1,2]},{"text":"The rule applies only in extreme cases and excludes ordinary frustration and research.","label":"Confirmed","sources":[1]},{"text":"The election section is renamed “Do Not Undermine Democratic Processes”.","label":"Confirmed","sources":[1,2]},{"text":"The blanket ban on personalised campaign targeting was removed.","label":"Confirmed","sources":[1]},{"text":"Weapons software, arming drones and tracking people without consent are explicitly banned.","label":"Confirmed","sources":[1]},{"text":"Whether breaking the abuse rule can lead to account suspension.","label":"Not known","sources":[]}],"sources":[{"title":"Anthropic: 2026 usage policy update","url":"https://www.anthropic.com/news/2026-usage-policy-update","type":"official"},{"title":"Anthropic: Usage Policy (effective 12 November 2026)","url":"https://www.anthropic.com/legal/aup","type":"official"},{"title":"MediaNama: Anthropic blocks abusive behaviour towards Claude","url":"https://www.medianama.com/2026/10/223-anthropic-abuse-claude-ban-campaign-targeting/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/anthropic-usage-policy-november-12-changes.json"},{"url":"https://anawords.com/story/google-gemini-agent-coworker-workspace-accounts","title":"Google’s new Gemini agent can create AI coworkers with their own email and calendar","summary":"Announced at Gemini at Work 2026, coworker agents get a Workspace account, Drive storage and a place in the company directory. Pricing for wider use is not…","section":"AI Agents","published":"2026-10-11T19:48:50+05:30","updated":"2026-10-11T19:48:50+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/google-gemini-agent-coworker-workspace-accounts","facts":[{"text":"Google announced the Gemini agent at Gemini at Work 2026 on 8 October.","label":"Confirmed","sources":[1]},{"text":"Coworker agents get their own Workspace account, email, calendar, Drive and directory entry.","label":"Confirmed","sources":[1]},{"text":"Coworker agents act under their own identity and see only what is shared with them.","label":"Confirmed","sources":[1]},{"text":"Admins can set hard spending limits that pause agents.","label":"Confirmed","sources":[1]},{"text":"The agent uses Gemini models and Anthropic’s Claude models today.","label":"Confirmed","sources":[1]},{"text":"Private preview for select customers at no extra charge for Gemini Enterprise.","label":"Unconfirmed","sources":[2]},{"text":"General availability date and full pricing.","label":"Not known","sources":[]}],"sources":[{"title":"Google Cloud Blog: Welcome to Gemini at Work 2026","url":"https://cloud.google.com/blog/products/ai-machine-learning/welcome-to-gemini-at-work-2026","type":"official"},{"title":"VentureBeat: Google Cloud unveils persistent Gemini agents","url":"https://venturebeat.com/orchestration/google-cloud-unveils-persistent-gemini-agents-for-long-running-tasks-and-they-get-their-own-gmail-calendar-and-drive-storage","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/google-gemini-agent-coworker-workspace-accounts.json"},{"url":"https://anawords.com/story/mistral-large-4-public-preview","title":"Mistral Large 4 enters preview with 1 trillion parameters, open weights due this month","summary":"The French lab’s biggest model is on its API now at $1.36 per million input tokens. Its benchmark scores are Mistral’s own and not yet checked independently.","section":"Models & Launches","published":"2026-10-11T19:47:50+05:30","updated":"2026-10-11T19:47:50+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/mistral-large-4-public-preview","facts":[{"text":"Public API preview opened on 6 October 2026.","label":"Confirmed","sources":[1]},{"text":"Mixture-of-experts model with 1 trillion total and 52 billion active parameters.","label":"Confirmed","sources":[1]},{"text":"List price is $1.36 per 1M input tokens and $4.18 per 1M output tokens.","label":"Confirmed","sources":[1]},{"text":"Open weights are due by the end of October, Mistral says.","label":"Confirmed","sources":[1]},{"text":"Trained on 3,800 NVIDIA Grace Blackwell GPUs in Europe.","label":"Company claim","sources":[1]},{"text":"Scores 61.7% on DeepSWE v1.1.","label":"Company claim","sources":[2]},{"text":"A two-week launch discount brings prices to $0.68 and $2.09.","label":"Unconfirmed","sources":[2]},{"text":"The licence for the open weights.","label":"Not known","sources":[]}],"sources":[{"title":"Mistral AI: Introducing Mistral Large 4","url":"https://mistral.ai/news/mistral-large-4/","type":"official"},{"title":"eWeek: Mistral Large 4 enters preview","url":"https://www.eweek.com/news/mistral-large-4-public-preview-enterprise-ai/","type":"outlet"},{"title":"VentureBeat: Mistral debuts Large 4","url":"https://venturebeat.com/technology/mistral-debuts-large-4-le-chonk-a-1-trillion-parameter-text-output-model-with-high-benchmarks-planned-for-open-weights-release","type":"outlet"}],"model_card":{"name":"Mistral Large 4","lab":"Mistral AI","released":"6 Oct 2026 (preview)","context":"Not stated by Mistral","input_price":"$1.36","output_price":"$4.18","india":"Not stated","licence":"Open weights due end of Oct","replaces":"Mistral Large 3"},"data":"https://anawords.com/story/mistral-large-4-public-preview.json"},{"url":"https://anawords.com/story/india-ai-regulation-consultation-paper-vaishnaw","title":"India will publish an AI regulation paper within a month, says IT Minister Vaishnaw","summary":"Safety and deepfakes top the list. The minister said on 8 October that the paper will be the base for India’s AI regulation.","section":"AI in India","published":"2026-10-11T19:46:50+05:30","updated":"2026-10-11T19:46:50+05:30","version":1,"checked":"sources checked by AI, approved by editor","research_note":"https://anawords.com/research/india-ai-regulation-consultation-paper-vaishnaw","facts":[{"text":"The government will release a consultation paper on AI regulation within a month.","label":"Reported","sources":[1,2,3]},{"text":"Safety will be the paper’s main focus.","label":"Reported","sources":[1,2]},{"text":"Deepfakes, user harm, cybersecurity and psychological harm are on the list.","label":"Reported","sources":[1,2]},{"text":"The minister spoke at the World Development Report release event on 8 October.","label":"Reported","sources":[1,2]},{"text":"He called for a techno-legal approach, with industry taking primary responsibility.","label":"Reported","sources":[1,2]},{"text":"The exact release date of the paper.","label":"Not known","sources":[]},{"text":"The legal form of the rules and how they will be enforced.","label":"Not known","sources":[]}],"sources":[{"title":"ThePrint (PTI): AI regulations: Consultation paper in a month","url":"https://theprint.in/economy/ai-regulations-consultation-paper-in-a-month-safety-deepfakes-among-key-focus-areas-says-vaishnaw/3065421/","type":"outlet"},{"title":"Business Today: “Time has come”: Vaishnaw says AI must be regulated","url":"https://www.businesstoday.in/technology/artificial-intelligence/story/time-has-come-ashwini-vaishnaw-says-ai-must-be-regulated-centre-to-release-consultation-paper-560439-2026-10-08","type":"outlet"},{"title":"Organiser: India moves to rein in AI risks","url":"https://organiser.org/2026/10/08/385395/bharat/india-moves-to-rein-in-ai-risks-government-to-unveil-regulation-blueprint-within-a-month/","type":"outlet"}],"model_card":null,"data":"https://anawords.com/story/india-ai-regulation-consultation-paper-vaishnaw.json"}]}