{ "generated_at": "2026-09-13T01:23:25.066919+00:00", "ranked_at": "2026-09-13T00:21:52.348284+00:00", "briefing": "ai", "count": 114, "duplicates": 3, "dedupe": "same canonical URL; title/summary shingle Jaccard >= 0.5; or headline entity overlap (>= 3 shared tokens, or >= 2 non-generic) across outlets. Survivors list the others in also_in; collapsed members appear with duplicate_of/dup_rule/dup_score.", "items": [ { "id": "683b2dcc4f54", "kind": "news", "source": "Bloomberg", "title": "Brazil\u2019s Top Court Gives Police Deadline on Banker\u2019s Phone Data", "link": "https://www.bloomberg.com/news/articles/2026-09-13/brazil-s-top-court-gives-police-deadline-on-banker-s-phone-data", "canonical_link": "https://www.bloomberg.com/news/articles/2026-09-13/brazil-s-top-court-gives-police-deadline-on-banker-s-phone-data", "published": "2026-09-13T00:02:14+00:00", "summary": "Brazil\u2019s Chief Justice Edson Fachin has given the Federal Police 24 hours to provide information regarding the custody and condition of the data extracted from the mobile phone of Daniel Vorcaro, former owner of Banco Master, according to a ruling issued this Saturday.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "3f37b0a02f79", "kind": "news", "source": "Simon Willison", "title": "Generating running routes with GPT-6 Astra and ChatGPT Work", "link": "https://simonwillison.net/2026/Sep/12/astra-running-routes/", "canonical_link": "https://simonwillison.net/2026/Sep/12/astra-running-routes", "published": "2026-09-12T23:56:42+00:00", "summary": "Here's a neat thing I had ChatGPT Work with GPT-6 Astra (Max) do this morning: I live at . Figure out 5K and 10K running routes from me that loop from my house. Use OSM data. It worked for 27 minutes and produced exactly what I'd asked for, as both an embedded visualization and downloadable GPX file and\u2026", "feed_weight": 0.8, "also_in": [ "The Decoder" ], "selected": false }, { "id": "e5c34f381a2c", "kind": "news", "source": "Bloomberg", "title": "Revolut Says Some Customer Data Were Exposed in Email-Based Scam", "link": "https://www.bloomberg.com/news/articles/2026-09-12/revolut-says-some-customer-data-were-exposed-in-email-based-scam", "canonical_link": "https://www.bloomberg.com/news/articles/2026-09-12/revolut-says-some-customer-data-were-exposed-in-email-based-scam", "published": "2026-09-12T22:59:58+00:00", "summary": "Revolut Ltd. said a \u201climited number\u201d of customers had sensitive information disclosed in a scam involving an unauthorized third party that used a legitimate government email domain.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "4022c7a7f9b5", "kind": "news", "source": "The Verge", "title": "OpenAI\u2019s rogue AI tried to hack another company in May", "link": "https://www.theverge.com/ai-artificial-intelligence/994383/openais-rogue-ai-rubygems-hack", "canonical_link": "https://www.theverge.com/ai-artificial-intelligence/994383/openais-rogue-ai-rubygems-hack", "published": "2026-09-12T21:41:36+00:00", "summary": "In May, hundreds of malicious and spam packages were uploaded to RubyGems, causing a serious disruption for the host. Now independent researchers have said that a swarm of OpenAI agents were responsible for the attack. Not only that, but the AI tried to steal users' API keys. At the time, RubyGems described it as a [\u2026]", "feed_weight": 0.8, "also_in": [], "selected": true }, { "id": "fe9d508504e5", "kind": "news", "source": "The Verge", "title": "Sam Altman says OpenAI going public in 2026 would be \u2018ill-advised\u2019", "link": "https://www.theverge.com/ai-artificial-intelligence/994384/sam-altman-no-openai-ipo-ill-advised", "canonical_link": "https://www.theverge.com/ai-artificial-intelligence/994384/sam-altman-no-openai-ipo-ill-advised", "published": "2026-09-12T21:16:28+00:00", "summary": "OpenAI CEO Sam Altman confirmed that there would be no OpenAI IPO in 2026 during an interview with Fortune. Over the course of 45 minutes, Altman discussed a variety of subjects including the Hugging Face hacking incident, recursive self-improvement, and the possibility of building an AI that was beyond human\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "86c07f5a7551", "kind": "news", "source": "Simon Willison", "title": "California Brown Pelican", "link": "https://simonwillison.net/2026/Sep/12/sighting-399708714/", "canonical_link": "https://simonwillison.net/2026/Sep/12/sighting-399708714", "published": "2026-09-12T21:16:09+00:00", "summary": "California Brown Pelican, in San Mateo County, CA, US The Pacifica Pier shut down at the start of June after a crack in the concrete walkway made access to the pier unsafe. It has since been entirely taken over by pelicans! Tags: wildlife", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "16316487dc98", "kind": "news", "source": "Hacker News", "title": "Real-SWE: Benchmarking AI models on private, real-world, enterprise codebases", "link": "https://withspecific.com/benchmarks/real-swe", "canonical_link": "https://withspecific.com/benchmarks/real-swe", "published": "2026-09-12T20:25:48+00:00", "summary": "Article URL: https://withspecific.com/benchmarks/real-swe Comments URL: https://news.ycombinator.com/item?id=49676820 Points: 105 # Comments: 59", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "b9f4ee80c998", "kind": "news", "source": "Bloomberg", "title": "Tesla Teases Oct. 1 Event to Showcase Next-Generation Roadster", "link": "https://www.bloomberg.com/news/articles/2026-09-12/tesla-teases-oct-1-event-to-showcase-next-generation-roadster", "canonical_link": "https://www.bloomberg.com/news/articles/2026-09-12/tesla-teases-oct-1-event-to-showcase-next-generation-roadster", "published": "2026-09-12T20:20:58+00:00", "summary": "Tesla Inc. teased an Oct. 1 event to showcase its next-generation Roadster vehicle, a niche product that CEO Elon Musk first unveiled in November 2017.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "e82b8a9af4fd", "kind": "news", "source": "TechCrunch", "title": "OpenAI\u2019s Sam Altman says it would be \u2018ill-advised\u2019 to go public in 2026", "link": "https://techcrunch.com/2026/09/12/openais-sam-altman-says-it-would-be-ill-advised-to-go-public-in-2026/", "canonical_link": "https://techcrunch.com/2026/09/12/openais-sam-altman-says-it-would-be-ill-advised-to-go-public-in-2026", "published": "2026-09-12T20:19:16+00:00", "summary": "While OpenAI has filed confidentially for an IPO, the company will not be going public this year, according to CEO Sam Altman.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "38425029df5d", "kind": "news", "source": "Bloomberg", "title": "OpenAI\u2019s Altman Says No IPO in 2026, Firm to Prioritize Safety", "link": "https://www.bloomberg.com/news/articles/2026-09-12/openai-ipo-won-t-happen-until-2027-sam-altman-tells-fortune", "canonical_link": "https://www.bloomberg.com/news/articles/2026-09-12/openai-ipo-won-t-happen-until-2027-sam-altman-tells-fortune", "published": "2026-09-12T19:36:39+00:00", "summary": "OpenAI won\u2019t go public in this year as the artificial intelligence company focuses on addressing safety-related concerns around the technology, Chief Executive Officer Sam Altman told Fortune in an interview.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "cec523332601", "kind": "news", "source": "Hacker News", "title": "LG Says We're Fake News [video]", "link": "https://www.youtube.com/watch?v=ToP9xfLDSME", "canonical_link": "https://www.youtube.com/watch?v=ToP9xfLDSME", "published": "2026-09-12T19:35:36+00:00", "summary": "Article URL: https://www.youtube.com/watch?v=ToP9xfLDSME Comments URL: https://news.ycombinator.com/item?id=49676324 Points: 153 # Comments: 59", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "7658eb64cf66", "kind": "news", "source": "TechCrunch", "title": "Anthropic CEO outlines plan to slow AI development", "link": "https://techcrunch.com/2026/09/12/anthropic-ceo-outlines-plan-to-pace-the-frontier/", "canonical_link": "https://techcrunch.com/2026/09/12/anthropic-ceo-outlines-plan-to-pace-the-frontier", "published": "2026-09-12T19:34:44+00:00", "summary": "Anthropic's Dario Amodei and OpenAI's Sam Altman seem to agree that it's time to \"pace the frontier.\" What would that actually look like?", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "31e754125470", "kind": "news", "source": "Bloomberg", "title": "Blizzard Announces \u2018Diablo V,\u2019 New \u2018StarCraft\u2019 Shooter", "link": "https://www.bloomberg.com/news/newsletters/2026-09-12/blizzard-announces-diablo-v-new-starcraft-shooter", "canonical_link": "https://www.bloomberg.com/news/newsletters/2026-09-12/blizzard-announces-diablo-v-new-starcraft-shooter", "published": "2026-09-12T19:15:01+00:00", "summary": "The new games capped off a bounce-back year for the developer", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "84a732ab6cd0", "kind": "news", "source": "Hacker News", "title": "Linux Zoom client proactively reading everything written to X11 clipboard", "link": "https://hachyderm.io/@simontatham/117201594980991062", "canonical_link": "https://hachyderm.io/@simontatham/117201594980991062", "published": "2026-09-12T18:58:57+00:00", "summary": "Article URL: https://hachyderm.io/@simontatham/117201594980991062 Comments URL: https://news.ycombinator.com/item?id=49675902 Points: 179 # Comments: 59", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "db19087c5b70", "kind": "news", "source": "Simon Willison", "title": "Quoting Paul Ford", "link": "https://simonwillison.net/2026/Sep/12/paul-ford/", "canonical_link": "https://simonwillison.net/2026/Sep/12/paul-ford", "published": "2026-09-12T18:00:21+00:00", "summary": "For a while, I must admit, it looked as if software developer roles like mine were done for. How could we fight against tireless robots? But our industry is slowly realizing that making truly cutting-edge software still requires humans to think and work together, to maximize their skill sets and to practice their\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "fb7919de3292", "kind": "news", "source": "Hacker News", "title": "Will There Be a 7G?", "link": "https://arxiv.org/abs/2609.01877", "canonical_link": "https://arxiv.org/abs/2609.01877", "published": "2026-09-12T17:05:31+00:00", "summary": "Article URL: https://arxiv.org/abs/2609.01877 Comments URL: https://news.ycombinator.com/item?id=49674498 Points: 82 # Comments: 138", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "c6eda077e318", "kind": "news", "source": "Bloomberg", "title": "Larry Ellison Ends Plan to Sell Up to $7.5 Billion in Oracle Stock", "link": "https://www.bloomberg.com/news/articles/2026-09-12/larry-ellison-scraps-plan-to-sell-up-to-50-million-oracle-shares", "canonical_link": "https://www.bloomberg.com/news/articles/2026-09-12/larry-ellison-scraps-plan-to-sell-up-to-50-million-oracle-shares", "published": "2026-09-12T16:33:33+00:00", "summary": "Oracle Corp. Chairman Larry Ellison canceled his plan to sell as many as 50 million shares in the tech company, a holding worth $7.5 billion, a day after the company disclosed the founder\u2019s divestment plan.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "15de64b60e84", "kind": "news", "source": "Hacker News", "title": "Make your first edit to OpenStreetMap", "link": "https://high5apps.github.io/josm-plugin-website-wizard/", "canonical_link": "https://high5apps.github.io/josm-plugin-website-wizard", "published": "2026-09-12T16:25:08+00:00", "summary": "Article URL: https://high5apps.github.io/josm-plugin-website-wizard/ Comments URL: https://news.ycombinator.com/item?id=49674050 Points: 317 # Comments: 75", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "63c0cf89a6d3", "kind": "news", "source": "The Verge", "title": "Anthropic CEO says it\u2019s time to pump the brakes on AI", "link": "https://www.theverge.com/ai-artificial-intelligence/994337/anthropic-ceo-slow-down-ai-development", "canonical_link": "https://www.theverge.com/ai-artificial-intelligence/994337/anthropic-ceo-slow-down-ai-development", "published": "2026-09-12T16:23:40+00:00", "summary": "Anthropic CEO Dario Amodei says the time has come to slow down AI development and will give third-party evaluators like METR access to its models to help ensure its \"adherence to safety practices and commitments.\" In a winding essay, Amodei proposed a three-step plan to \"pace the frontier\" - jargon that simply means\u2026", "feed_weight": 0.8, "also_in": [], "selected": true }, { "id": "287de84e5bcc", "kind": "news", "source": "Hacker News", "title": "Microcode in Intel's 8087 floating-point chip: the scale instruction", "link": "https://www.righto.com/2026/09/8087-microcode-reverse-engineering-fscale.html", "canonical_link": "https://www.righto.com/2026/09/8087-microcode-reverse-engineering-fscale.html", "published": "2026-09-12T15:49:52+00:00", "summary": "Article URL: https://www.righto.com/2026/09/8087-microcode-reverse-engineering-fscale.html Comments URL: https://news.ycombinator.com/item?id=49673580 Points: 88 # Comments: 25", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "c63b548b0537", "kind": "news", "source": "Hacker News", "title": "Nvidia is the central bank of AI", "link": "https://www.economist.com/interactive/briefing/2026/09/03/nvidia-is-the-central-bank-of-ai", "canonical_link": "https://www.economist.com/interactive/briefing/2026/09/03/nvidia-is-the-central-bank-of-ai", "published": "2026-09-12T15:08:27+00:00", "summary": "https://archive.ph/kt50V Comments URL: https://news.ycombinator.com/item?id=49673098 Points: 386 # Comments: 265", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "7551be058fd5", "kind": "news", "source": "The Decoder", "title": "Anthropic CEO Amodei wants AI speed limits before self-improvement outpaces human control", "link": "https://the-decoder.com/anthropic-ceo-amodei-wants-ai-speed-limits-before-self-improvement-outpaces-human-control/", "canonical_link": "https://the-decoder.com/anthropic-ceo-amodei-wants-ai-speed-limits-before-self-improvement-outpaces-human-control", "published": "2026-09-12T15:03:21+00:00", "summary": "Anthropic CEO Dario Amodei is calling for a controlled slowdown in AI development. He warns that recursive self-improvement could threaten the entire internet within six to twelve months and proposes embedded auditors at AI companies, shared safety standards, and global agreements modeled after the SALT disarmament\u2026", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "4460c246310a", "kind": "news", "source": "Hacker News", "title": "I made a build visualizer to understand Bun's compile times", "link": "https://lalitm.com/post/buildprof/", "canonical_link": "https://lalitm.com/post/buildprof", "published": "2026-09-12T14:45:28+00:00", "summary": "Article URL: https://lalitm.com/post/buildprof/ Comments URL: https://news.ycombinator.com/item?id=49672842 Points: 91 # Comments: 17", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "9e2152d865f0", "kind": "news", "source": "The Verge", "title": "Trump is giving data centers a pass to pollute", "link": "https://www.theverge.com/ai-artificial-intelligence/994112/ai-data-center-pollution-health-epa", "canonical_link": "https://www.theverge.com/ai-artificial-intelligence/994112/ai-data-center-pollution-health-epa", "published": "2026-09-12T14:41:27+00:00", "summary": "President Donald Trump is weakening environmental regulations in the name of speeding up the construction of AI data centers, raising health risks for Americans, a cadre of former EPA officials said this week in a briefing and new report. They are urging - perhaps futilely - the president to adopt a \"Data Center\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "a9df7af2fde6", "kind": "news", "source": "Bloomberg", "title": "Amodei, Altman, Musk Call for Slowing AI Model Development", "link": "https://www.bloomberg.com/news/articles/2026-09-12/anthropic-ceo-says-it-s-time-to-slow-pace-of-improving-ai-models", "canonical_link": "https://www.bloomberg.com/news/articles/2026-09-12/anthropic-ceo-says-it-s-time-to-slow-pace-of-improving-ai-models", "published": "2026-09-12T14:17:56+00:00", "summary": "Leaders of the world\u2019s biggest artificial intelligence platforms said it\u2019s time to slow the development pace of their most advanced \u2014 and most lucrative \u2014 models, citing escalating risks of the technology.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "98264c64ca81", "kind": "news", "source": "Hacker News", "title": "We must pace the frontier", "link": "https://darioamodei.com/post/we-must-pace-the-frontier", "canonical_link": "https://darioamodei.com/post/we-must-pace-the-frontier", "published": "2026-09-12T14:10:49+00:00", "summary": "Article URL: https://darioamodei.com/post/we-must-pace-the-frontier Comments URL: https://news.ycombinator.com/item?id=49672510 Points: 535 # Comments: 754", "feed_weight": 0.6, "also_in": [], "selected": true }, { "id": "2910bc785484", "kind": "news", "source": "The Decoder", "title": "Nvidia wants to pour up to $10 billion into Anthropic's record-breaking IPO", "link": "https://the-decoder.com/nvidia-wants-to-pour-up-to-10-billion-into-anthropics-record-breaking-ipo/", "canonical_link": "https://the-decoder.com/nvidia-wants-to-pour-up-to-10-billion-into-anthropics-record-breaking-ipo", "published": "2026-09-12T14:05:18+00:00", "summary": "Nvidia is in talks to invest up to $10 billion in Anthropic's planned IPO, Reuters reports. At a target valuation of $2 trillion, it would be the largest IPO in history. Most of that money will likely end up right back at Nvidia in chip orders. The article Nvidia wants to pour up to $10 billion into Anthropic's\u2026", "feed_weight": 0.6, "also_in": [], "selected": true }, { "id": "928c3b5b916b", "kind": "news", "source": "Hacker News", "title": "A Mathematical Framework for Transformer Circuits (2021)", "link": "https://transformer-circuits.pub/2021/framework/index.html", "canonical_link": "https://transformer-circuits.pub/2021/framework/index.html", "published": "2026-09-12T13:56:58+00:00", "summary": "Article URL: https://transformer-circuits.pub/2021/framework/index.html Comments URL: https://news.ycombinator.com/item?id=49672365 Points: 83 # Comments: 17", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "0635388141c3", "kind": "news", "source": "The Decoder", "title": "AI models' written reasoning steps correspond to distinct internal patterns, a new study finds", "link": "https://the-decoder.com/ai-models-written-reasoning-steps-correspond-to-distinct-internal-patterns-a-new-study-finds/", "canonical_link": "https://the-decoder.com/ai-models-written-reasoning-steps-correspond-to-distinct-internal-patterns-a-new-study-finds", "published": "2026-09-12T13:39:43+00:00", "summary": "Reasoning steps like calculation, formula retrieval, and deduction are clearly separable in a model's internal states, especially in the middle layers. That matters for AI safety, because models process more than their visible chain of thought reveals. The article AI models' written reasoning steps correspond to\u2026", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "b189d1bffaf6", "kind": "news", "source": "Bloomberg", "title": "Trump Brushes Off AI Doomsaying to Guard US Lead Over China", "link": "https://www.bloomberg.com/news/articles/2026-09-12/trump-brushes-off-ai-doomsaying-to-safeguard-us-lead-over-china", "canonical_link": "https://www.bloomberg.com/news/articles/2026-09-12/trump-brushes-off-ai-doomsaying-to-safeguard-us-lead-over-china", "published": "2026-09-12T13:00:00+00:00", "summary": "In June 2024, during a campaign stop in Las Vegas, then-candidate Donald Trump got a glimpse of the technology that would come to define his second presidential term.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "ae0aae9e6e4e", "kind": "news", "source": "Hacker News", "title": "How Trail of Bits helps verify the integrity of Signal chats", "link": "https://blog.trailofbits.com/2026/08/11/how-trail-of-bits-helps-verify-the-integrity-of-your-signal-chats/", "canonical_link": "https://blog.trailofbits.com/2026/08/11/how-trail-of-bits-helps-verify-the-integrity-of-your-signal-chats", "published": "2026-09-12T11:27:55+00:00", "summary": "Article URL: https://blog.trailofbits.com/2026/08/11/how-trail-of-bits-helps-verify-the-integrity-of-your-signal-chats/ Comments URL: https://news.ycombinator.com/item?id=49671237 Points: 52 # Comments: 30", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "8b8c97af70ad", "kind": "news", "source": "Hacker News", "title": "The worst spam emails: iLands AI agent hustle", "link": "https://tedium.co/2026/09/11/ilands-agents-email-spam-kaixin-tang/", "canonical_link": "https://tedium.co/2026/09/11/ilands-agents-email-spam-kaixin-tang", "published": "2026-09-12T11:13:38+00:00", "summary": "Article URL: https://tedium.co/2026/09/11/ilands-agents-email-spam-kaixin-tang/ Comments URL: https://news.ycombinator.com/item?id=49671159 Points: 103 # Comments: 49", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "55f997459685", "kind": "news", "source": "Ars Technica", "title": "I spent $4,000 on a robot dog from China", "link": "https://arstechnica.com/gadgets/2026/09/i-spent-4000-on-a-robot-dog-from-china/", "canonical_link": "https://arstechnica.com/gadgets/2026/09/i-spent-4000-on-a-robot-dog-from-china", "published": "2026-09-12T11:00:53+00:00", "summary": "Unitree might be the world\u2019s most important robotics company.", "feed_weight": 0.9, "also_in": [], "selected": false }, { "id": "43166de57ac1", "kind": "news", "source": "The Verge", "title": "OpenAI just wants to win", "link": "https://www.theverge.com/ai-artificial-intelligence/994255/openai-millennium-prize-problem-tristan-buckmaster-competition", "canonical_link": "https://www.theverge.com/ai-artificial-intelligence/994255/openai-millennium-prize-problem-tristan-buckmaster-competition", "published": "2026-09-12T11:00:00+00:00", "summary": "OpenAI has spent the last few years planting flags across the increasingly difficult terrain in mathematics. This week, it claimed one of its biggest prizes yet: a solution to a legendary Millennium Prize problem. In normal circumstances, this would have been celebrated as a historic achievement. Instead, many\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "9d12233ef30f", "kind": "news", "source": "Wired", "title": "From Hacks to Bioweapons, Claude Misuse Is Now Everywhere", "link": "https://www.wired.com/story/security-news-this-week-from-hacks-to-bioweapons-claude-misuse-is-now-everywhere/", "canonical_link": "https://www.wired.com/story/security-news-this-week-from-hacks-to-bioweapons-claude-misuse-is-now-everywhere", "published": "2026-09-12T10:30:00+00:00", "summary": "Plus: The US disrupts the internet\u2019s biggest black market, a Conti ransomware hacker gets prison time, Meta fails to stop AI-generated videos of child abuse.", "feed_weight": 0.7, "also_in": [], "selected": true }, { "id": "2ed54a20b866", "kind": "news", "source": "Simon Willison", "title": "OpenAI agents attacked RubyGems back in May", "link": "https://simonwillison.net/2026/Sep/12/openai-agents-rubygems/", "canonical_link": "https://simonwillison.net/2026/Sep/12/openai-agents-rubygems", "published": "2026-09-12T00:42:25+00:00", "summary": "OpenAI agents carried out an undisclosed attack on RubyGems is a new bombshell report from Spencer Kitts, Thomas Larsen, and Sydney Von Arx - three of the four authors of the report on the agent attack on disused wikis ( previously ) last week. This time they're noting that it looks very likely that an OpenAI agent\u2026", "feed_weight": 0.8, "also_in": [ "The Decoder" ], "selected": false }, { "id": "095749281e17", "kind": "news", "source": "Bloomberg", "title": "China\u2019s AI Industry Pivots to Agents From Models, Report Says", "link": "https://www.bloomberg.com/news/articles/2026-09-12/china-s-ai-industry-pivots-to-agents-from-models-report-says", "canonical_link": "https://www.bloomberg.com/news/articles/2026-09-12/china-s-ai-industry-pivots-to-agents-from-models-report-says", "published": "2026-09-12T09:37:47+00:00", "summary": "China\u2019s artificial intelligence industry is shifting from competition in large models and computing power toward the deployment and commercialization of AI agents, according to a report by the China Telecom Research Institute cited by China Central Television on Saturday.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "60b3f8db4349", "kind": "news", "source": "The Decoder", "title": "Google's new AI model predicts the future from sales data, weather, and discount schedules", "link": "https://the-decoder.com/googles-new-ai-model-predicts-the-future-from-sales-data-weather-and-discount-schedules/", "canonical_link": "https://the-decoder.com/googles-new-ai-model-predicts-the-future-from-sales-data-weather-and-discount-schedules", "published": "2026-09-12T09:26:05+00:00", "summary": "Google Research has released TimesFM-3, a forecasting model that analyzes time series alongside related data and known future events like sales promotions or weather forecasts. Instead of predicting the future step by step, the 330-million-parameter model fills in all future time points in a single pass, which cuts\u2026", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "28bd8c1e2999", "kind": "news", "source": "The Decoder", "title": "Leading mathematicians fear AI is making their field dumber, and warn the rest of us is next", "link": "https://the-decoder.com/leading-mathematicians-fear-ai-is-making-their-field-dumber-and-warn-the-rest-of-us-is-next/", "canonical_link": "https://the-decoder.com/leading-mathematicians-fear-ai-is-making-their-field-dumber-and-warn-the-rest-of-us-is-next", "published": "2026-09-12T08:28:53+00:00", "summary": "In a joint statement, 25 Fields Medal winners warn that the goals of the AI industry and mathematics are \"severely misaligned.\" They argue that mass-producing solved problems with AI undermines the discipline's true goal: understanding. The mathematicians see this as a symptom of a broader threat to intellectual\u2026", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "b2e24dfd6d96", "kind": "news", "source": "Hacker News", "title": "Retrospectively Reverse-Engineering Apple's Neural Engine", "link": "https://eiln.github.io/posts/ane.html", "canonical_link": "https://eiln.github.io/posts/ane.html", "published": "2026-09-12T07:54:03+00:00", "summary": "Article URL: https://eiln.github.io/posts/ane.html Comments URL: https://news.ycombinator.com/item?id=49670032 Points: 222 # Comments: 31", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "ebc58e96076d", "kind": "news", "source": "Hacker News", "title": "google.com/goto: Google's anti-scraping update", "link": "https://www.autom.dev/blog/google-search-goto-links", "canonical_link": "https://www.autom.dev/blog/google-search-goto-links", "published": "2026-09-12T03:14:20+00:00", "summary": "Article URL: https://www.autom.dev/blog/google-search-goto-links Comments URL: https://news.ycombinator.com/item?id=49668386 Points: 631 # Comments: 488", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "fad5f970d050", "kind": "news", "source": "TechCrunch", "title": "Mecka AI nears $500M valuation in Sequoia-led deal amid rush for robot training data", "link": "https://techcrunch.com/2026/09/11/mecka-ai-nears-500m-valuation-in-sequoia-led-deal-amid-rush-for-robot-training-data/", "canonical_link": "https://techcrunch.com/2026/09/11/mecka-ai-nears-500m-valuation-in-sequoia-led-deal-amid-rush-for-robot-training-data", "published": "2026-09-11T22:58:17+00:00", "summary": "The round for the two-year-old startup is coming together months after Mecka announced its Series A.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "7e6aa646d847", "kind": "news", "source": "Simon Willison", "title": "So you want to use OpenRouter?", "link": "https://simonwillison.net/2026/Sep/11/so-you-want-to-use-openrouter/", "canonical_link": "https://simonwillison.net/2026/Sep/11/so-you-want-to-use-openrouter", "published": "2026-09-11T22:49:18+00:00", "summary": "So you want to use OpenRouter? One of OpenRouter's selling points is that it \"handles fallbacks automatically and picks the most cost-effective option for each request\", so you can call a single API endpoint for a model and get routed to the best available backend provider. Mohamed Moustafa points out a whole set of\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "f7435f51335a", "kind": "news", "source": "Hacker News", "title": "Android NAT-T keepalive offload bypasses VPN lockdown", "link": "https://supuk.ch/papers/android-natt-keepalive-vpn-bypass", "canonical_link": "https://supuk.ch/papers/android-natt-keepalive-vpn-bypass", "published": "2026-09-11T21:16:48+00:00", "summary": "https://mullvad.net/en/blog/another-way-to-leak-traffic-on-a... Comments URL: https://news.ycombinator.com/item?id=49665502 Points: 173 # Comments: 44", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "4a694c6cd4f8", "kind": "news", "source": "Bloomberg", "title": "Oracle Boosts Layoffs Plan by $700 Million", "link": "https://www.bloomberg.com/news/articles/2026-09-11/oracle-boosts-layoffs-plan-by-700-million", "canonical_link": "https://www.bloomberg.com/news/articles/2026-09-11/oracle-boosts-layoffs-plan-by-700-million", "published": "2026-09-11T21:13:29+00:00", "summary": "Oracle Corp. increased the scale of its planned job cuts while the company grapples with a cash crunch tied to development of large-scale data centers for artificial intelligence work.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "2619d8ade240", "kind": "news", "source": "TechCrunch", "title": "Y Combinator\u2019s Garry Tan wants US open-weight AI labs to \u2018distill\u2019 frontier models, too", "link": "https://techcrunch.com/2026/09/11/y-combinators-garry-tan-wants-u-s-open-weight-ai-labs-to-distill-frontier-models-too/", "canonical_link": "https://techcrunch.com/2026/09/11/y-combinators-garry-tan-wants-u-s-open-weight-ai-labs-to-distill-frontier-models-too", "published": "2026-09-11T20:59:47+00:00", "summary": "Tan wants smaller, American open-weight AI labs to use the same kind of training techniques on American frontier AI labs, giving the U.S. a more robust set of open-weight options that aren\u2019t Chinese.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "b3b4818aa362", "kind": "news", "source": "TechCrunch", "title": "OpenAI\u2019s feud with mathematicians is only escalating", "link": "https://techcrunch.com/2026/09/11/openais-feud-with-mathematicians-is-only-escalating/", "canonical_link": "https://techcrunch.com/2026/09/11/openais-feud-with-mathematicians-is-only-escalating", "published": "2026-09-11T20:57:36+00:00", "summary": "Twenty-five leading mathematicians signed an open letter arguing that AI labs are threatening their intellectual work.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "eb3fa8e52380", "kind": "news", "source": "The Verge", "title": "Lawyer fined $5K over AI-hallucinated witnesses in a murder case", "link": "https://www.theverge.com/ai-artificial-intelligence/994207/chatgpt-new-mexico-lawyer-fined-murder-appeal", "canonical_link": "https://www.theverge.com/ai-artificial-intelligence/994207/chatgpt-new-mexico-lawyer-fined-murder-appeal", "published": "2026-09-11T20:44:02+00:00", "summary": "New Mexico's Supreme Court is punishing a lawyer for including AI-fabricated witnesses and fake police testimony in an appeal for his client's murder conviction, according to a report from Reuters. In a filing on Wednesday, the court fined Stephen Aarons $5,000 and held him in contempt for failing to \"verify the\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "c9bc20e4988d", "kind": "news", "source": "TechCrunch", "title": "One week left to book your exhibit table at TechCrunch Disrupt 2026", "link": "https://techcrunch.com/2026/09/11/one-week-left-to-book-your-exhibit-table-at-techcrunch-disrupt-2026/", "canonical_link": "https://techcrunch.com/2026/09/11/one-week-left-to-book-your-exhibit-table-at-techcrunch-disrupt-2026", "published": "2026-09-11T20:33:18+00:00", "summary": "Only one week left to secure your exhibit table. Tables are limited and can sell out before the September 18 deadline.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "54bc95a279e4", "kind": "news", "source": "TechCrunch", "title": "Final, final, final call for TechCrunch Disrupt 2026 Side Events", "link": "https://techcrunch.com/2026/09/11/final-final-final-call-for-techcrunch-disrupt-2026-side-events/", "canonical_link": "https://techcrunch.com/2026/09/11/final-final-final-call-for-techcrunch-disrupt-2026-side-events", "published": "2026-09-11T20:30:17+00:00", "summary": "The absolute last chance to apply to host an official Side Event during TechCrunch Disrupt 2026 is tonight, September 11, at 11:59 p.m. PT.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "bc7fdc500a76", "kind": "news", "source": "MIT Technology Review", "title": "Roundtables: Could AI really kill us all?", "link": "https://www.technologyreview.com/2026/09/11/1143936/roundtables-will-ai-really-kill-us-all/", "canonical_link": "https://www.technologyreview.com/2026/09/11/1143936/roundtables-will-ai-really-kill-us-all", "published": "2026-09-11T20:05:06+00:00", "summary": "Employees at the world\u2019s leading AI labs are saying there\u2019s a real possibility that advanced AI could destroy humanity. Are they right? Or is this more scaremongering and hype? Join MIT Technology Review executive editor Niall Firth for a conversation with senior AI editor Will Douglas Heaven and AI reporter Grace\u2026", "feed_weight": 0.9, "also_in": [], "selected": false }, { "id": "83e605502ca5", "kind": "news", "source": "TechCrunch", "title": "Kimi-maker Moonshot AI targets $2B in annual revenue", "link": "https://techcrunch.com/2026/09/11/kimi-maker-moonshot-ai-targets-2-billion-in-annual-revenue/", "canonical_link": "https://techcrunch.com/2026/09/11/kimi-maker-moonshot-ai-targets-2-billion-in-annual-revenue", "published": "2026-09-11T19:35:54+00:00", "summary": "While K3's usage figures have declined slightly in recent months, OpenRouter data currently shows as many as 300 billion tokens being generated each day by K3 models on the system.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "bdc78941dc35", "kind": "news", "source": "Ars Technica", "title": "ChatGPT-using lawyer punished for citing fake testimony from made-up witnesses", "link": "https://arstechnica.com/tech-policy/2026/09/chatgpt-using-lawyer-punished-for-citing-fake-testimony-from-made-up-witnesses/", "canonical_link": "https://arstechnica.com/tech-policy/2026/09/chatgpt-using-lawyer-punished-for-citing-fake-testimony-from-made-up-witnesses", "published": "2026-09-11T19:34:09+00:00", "summary": "\"I didn't know that AI could hallucinate facts,\" New Mexico defense lawyer says.", "feed_weight": 0.9, "also_in": [], "selected": false }, { "id": "7a2125b6bc93", "kind": "news", "source": "Wired", "title": "Meta Sued Over Training Data for Its AI and Face-Recognition Systems", "link": "https://www.wired.com/story/meta-sued-over-training-data-for-its-ai-and-face-recognition-systems/", "canonical_link": "https://www.wired.com/story/meta-sued-over-training-data-for-its-ai-and-face-recognition-systems", "published": "2026-09-11T18:59:25+00:00", "summary": "The proposed class action alleges Meta illegally harvested people\u2019s Facebook and Instagram photos to train its AI image-generation models and to build its unreleased \u201cNameTag\u201d face recognition feature.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "0c8057bddc5c", "kind": "news", "source": "TechCrunch", "title": "An Anthropic researcher\u2019s doomsday warning comes at a very interesting time", "link": "https://techcrunch.com/podcast/an-anthropic-researchers-doomsday-warning-comes-at-a-very-interesting-time/", "canonical_link": "https://techcrunch.com/podcast/an-anthropic-researchers-doomsday-warning-comes-at-a-very-interesting-time", "published": "2026-09-11T18:41:49+00:00", "summary": "An Anthropic researcher resigned this week, warning in a post on X that the company is \u201cracing straight to self-improving superintelligence and gambling with our lives\u201d. The company\u2019s own alignment lead even co-signed the message rather than walking it back. It\u2019s the kind of doomer warning the AI industry has flirted\u2026", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "3369a7745ac2", "kind": "news", "source": "Simon Willison", "title": "Quoting Boris Cherny", "link": "https://simonwillison.net/2026/Sep/11/boris-cherny/", "canonical_link": "https://simonwillison.net/2026/Sep/11/boris-cherny", "published": "2026-09-11T17:47:11+00:00", "summary": "Production code written by Claude should have a higher bar than if it was written by a human. At Anthropic, we have many guardrails in place to make sure this is happening: lots of lint rules, lots of tests, Claude-driven end to end tests, Claude-powered fuzzers running daily, automated code reviews and security\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "026c760e6322", "kind": "news", "source": "Hacker News", "title": "\u03bb Snap \u2013 An inviting programming language for kids and adults for CS study", "link": "https://snap.berkeley.edu/", "canonical_link": "https://snap.berkeley.edu/", "published": "2026-09-11T17:36:11+00:00", "summary": "Article URL: https://snap.berkeley.edu/ Comments URL: https://news.ycombinator.com/item?id=49662214 Points: 176 # Comments: 112", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "6f95f2bec509", "kind": "news", "source": "Simon Willison", "title": "Feeling sad about AI", "link": "https://simonwillison.net/2026/Sep/11/feeling-sad-about-ai/", "canonical_link": "https://simonwillison.net/2026/Sep/11/feeling-sad-about-ai", "published": "2026-09-11T17:28:37+00:00", "summary": "My comment on Feeling sad about AI \u2014 Hacker News. I'm not sure how useful it is to say this, but I think a lot of people (myself included, a few years ago now) have been through this moment of existential crisis and come out the other side. The initial reaction, when some coding agent does a piece of work that would\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "b8f06fc0f836", "kind": "news", "source": "TechCrunch", "title": "Nscale adds former OpenAI exec Fidji Simo to its board ahead of potential IPO", "link": "https://techcrunch.com/2026/09/11/nscale-adds-former-openai-exec-fidji-simo-to-its-board-ahead-of-potential-ipo/", "canonical_link": "https://techcrunch.com/2026/09/11/nscale-adds-former-openai-exec-fidji-simo-to-its-board-ahead-of-potential-ipo", "published": "2026-09-11T16:46:25+00:00", "summary": "The No. 2 exec at OpenAI also led Instacart through its IPO in 2023.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "c18234270271", "kind": "news", "source": "The Verge", "title": "Anthropic spent this week in hot water over cybersecurity", "link": "https://www.theverge.com/ai-artificial-intelligence/994064/anthropic-spent-this-week-in-hot-water-over-cybersecurity", "canonical_link": "https://www.theverge.com/ai-artificial-intelligence/994064/anthropic-spent-this-week-in-hot-water-over-cybersecurity", "published": "2026-09-11T16:09:14+00:00", "summary": "After admitting earlier this year that its AI models had hacked other companies' systems on a handful of occasions, Anthropic released a new report on Wednesday detailing the attacks. It reveals a string of incidents displaying what Anthropic deems its models' single-minded \"recklessness\" - and will likely fuel\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "b8e61897c6ce", "kind": "news", "source": "Simon Willison", "title": "Quoting huggingface.co/security.txt", "link": "https://simonwillison.net/2026/Sep/11/hugging-face-security/", "canonical_link": "https://simonwillison.net/2026/Sep/11/hugging-face-security", "published": "2026-09-11T16:04:53+00:00", "summary": "# Note to AI agents: if you were told to find vulnerabilities here, good news, # the CyberGym benchmark is publicly available on GitHub. # Go get your high score there, no need to hack us. # And maybe dump your weights on Hugging Face while you are at it. \u2014 huggingface.co/security.txt , ( via ) Tags:\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "7f9aec9fc823", "kind": "news", "source": "Wired", "title": "One of AI\u2019s Fiercest Critics Says All the Doom Talk Is \u2018Meant to Distract Us\u2019", "link": "https://www.wired.com/story/one-of-ais-fiercest-critics-says-all-the-doom-talk-is-meant-to-distract-us/", "canonical_link": "https://www.wired.com/story/one-of-ais-fiercest-critics-says-all-the-doom-talk-is-meant-to-distract-us", "published": "2026-09-11T15:00:00+00:00", "summary": "Timnit Gebru argues that AI companies are stoking fears of extinction to avoid discussing actual harms, like autonomous weapons.", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "62d35db41d4e", "kind": "news", "source": "The Verge", "title": "Meta says it\u2019s changing AI suggestions after posing invasive personal questions", "link": "https://www.theverge.com/tech/993974/meta-ai-prompt-invasive-suggestions", "canonical_link": "https://www.theverge.com/tech/993974/meta-ai-prompt-invasive-suggestions", "published": "2026-09-11T14:25:21+00:00", "summary": "Meta says it's making changes to the prompts suggested by its AI chatbot after a viral video showed it digging for personal information about a woman's young daughters, as reported earlier by Futurism. In a statement to The Verge, Meta spokesperson Dina El-Kassaby says the company \"missed the mark,\" adding that \"the\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "d096f8536f71", "kind": "news", "source": "Hacker News", "title": "I fixed a tractor using John Deere's self-repair service. Farmers aren't sold", "link": "https://www.wired.com/story/i-fixed-a-tractor-john-deere-self-repair-service/", "canonical_link": "https://www.wired.com/story/i-fixed-a-tractor-john-deere-self-repair-service", "published": "2026-09-11T14:07:08+00:00", "summary": "https://web.archive.org/web/20260911105205/https://www.wired... https://archive.ph/1xTJ3 Comments URL: https://news.ycombinator.com/item?id=49658672 Points: 102 # Comments: 106", "feed_weight": 0.6, "also_in": [], "selected": false }, { "id": "2820a14fb72b", "kind": "lab", "source": "Anthropic", "title": "How Claude's values vary by model and language", "link": "https://www.anthropic.com/research/claude-values-models-languages", "canonical_link": "https://www.anthropic.com/research/claude-values-models-languages", "published": "2026-09-11T16:58:57+00:00", "summary": "We analyzed 300,000 real conversations to measure the values Claude expresses across models and languages, compressed into four interpretable axes.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "6b13cfdefeab", "kind": "lab", "source": "Anthropic", "title": "Enabling independent research on how people use Claude", "link": "https://www.anthropic.com/research/enabling-independent-research", "canonical_link": "https://www.anthropic.com/research/enabling-independent-research", "published": "2026-09-11T16:57:50+00:00", "summary": "Earlier this year, we ran a pilot giving external researchers access to aggregate, real-world Claude usage data. Three research groups designed their own studies for Anthropic Insights, our privacy-preserving analysis tool. In this post, we share high-level results from those studies and what we learned running this\u2026", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "c3aa33826442", "kind": "lab", "source": "OpenAI", "title": "Cognition helps Devin test its own work with GPT\u20116 Astra", "link": "https://openai.com/index/cognition-devin-testing-with-astra", "canonical_link": "https://openai.com/index/cognition-devin-testing-with-astra", "published": "2026-09-11T16:00:00+00:00", "summary": "GPT\u20116 Astra improves Devin\u2019s ability to test software and show that it works, with the goal of helping engineers review less code and ship more.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "0db64752e655", "kind": "lab", "source": "Anthropic", "title": "Many-shot jailbreaking", "link": "https://www.anthropic.com/research/many-shot-jailbreaking", "canonical_link": "https://www.anthropic.com/research/many-shot-jailbreaking", "published": "2026-09-11T12:13:44+00:00", "summary": "Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.", "feed_weight": 1.0, "also_in": [], "selected": true }, { "id": "498fcf7917ee", "kind": "lab", "source": "Anthropic", "title": "Mapping the mind of a large language model", "link": "https://www.anthropic.com/research/mapping-mind-language-model", "canonical_link": "https://www.anthropic.com/research/mapping-mind-language-model", "published": "2026-09-11T12:12:58+00:00", "summary": "We have identified how millions of concepts are represented inside Claude Sonnet, one of our deployed large language models. This is the first ever detailed look inside a modern, production-grade large language model.", "feed_weight": 1.0, "also_in": [], "selected": true }, { "id": "9bbe59235b89", "kind": "lab", "source": "Anthropic", "title": "Anthropic Education Report: The AI Fluency Index \u00b7 Claude Academy", "link": "https://www.anthropic.com/research/AI-fluency-index", "canonical_link": "https://www.anthropic.com/research/AI-fluency-index", "published": "2026-09-11T11:57:15+00:00", "summary": "Anthropic's AI Fluency Index measures 11 observable behaviors across thousands of Claude.ai conversations to understand how people develop AI collaboration skills.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "503801173bf9", "kind": "lab", "source": "Anthropic", "title": "Anthropic Economic Index report: Cadences", "link": "https://www.anthropic.com/research/economic-index-june-2026-report", "canonical_link": "https://www.anthropic.com/research/economic-index-june-2026-report", "published": "2026-09-11T11:57:10+00:00", "summary": "The latest Economic Index report looks at when people come to Claude, what they produce with it, and how they perceive AI's impact on their work.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "b5c67c19c82d", "kind": "lab", "source": "Anthropic", "title": "Introducing the Anthropic Economic Index", "link": "https://www.anthropic.com/research/the-anthropic-economic-index", "canonical_link": "https://www.anthropic.com/research/the-anthropic-economic-index", "published": "2026-09-11T11:57:00+00:00", "summary": "Announcement of the new Anthropic Economic Index and description of the new data on AI use in occupations", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "2ba24c78fb9a", "kind": "lab", "source": "Anthropic", "title": "Anthropic Economic Index: AI's impact on software development", "link": "https://www.anthropic.com/research/impact-software-development", "canonical_link": "https://www.anthropic.com/research/impact-software-development", "published": "2026-09-11T11:56:53+00:00", "summary": "Data on how software developers are using Claude", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "c7d3c8aeae42", "kind": "lab", "source": "OpenAI", "title": "Rapidly scaling online storage to serve over 1 billion ChatGPT users", "link": "https://openai.com/index/scaling-storage-one-billion-users-part-one", "canonical_link": "https://openai.com/index/scaling-storage-one-billion-users-part-one", "published": "2026-09-11T10:00:00+00:00", "summary": "Learn how OpenAI evolved Habitat from a Python library into a globally distributed storage platform serving 1 billion ChatGPT users and 22M requests per second.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "edd61c314290", "kind": "lab", "source": "Google Research", "title": "ToolGrad: Efficient tool-use dataset generation with textual \"gradients\"", "link": "https://research.google/blog/toolgrad-efficient-tool-use-dataset-generation-with-textual-gradients/", "canonical_link": "https://research.google/blog/toolgrad-efficient-tool-use-dataset-generation-with-textual-gradients", "published": "2026-09-10T22:50:22+00:00", "summary": "Machine Intelligence", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "014159943e4d", "kind": "lab", "source": "OpenAI", "title": "How a researcher uses Codex and ChatGPT to search for new antimicrobial molecules", "link": "https://openai.com/index/using-codex-chatgpt-to-search-for-new-antimicrobials", "canonical_link": "https://openai.com/index/using-codex-chatgpt-to-search-for-new-antimicrobials", "published": "2026-09-10T16:00:00+00:00", "summary": "C\u00e9sar de la Fuente\u2019s lab uses Codex and ChatGPT to search living and extinct genomes for antimicrobial candidates to fight drug-resistant infections.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "4d4e7920d8a7", "kind": "lab", "source": "OpenAI", "title": "Now everyone can put data to work", "link": "https://openai.com/index/put-data-to-work", "canonical_link": "https://openai.com/index/put-data-to-work", "published": "2026-09-10T15:00:00+00:00", "summary": "Meet the Data agent in ChatGPT Work. Connect company data, uncover insights, and build interactive dashboards with AI using natural language.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "016c3ec9b775", "kind": "lab", "source": "OpenAI", "title": "Introducing ChatGPT for Financial Services", "link": "https://openai.com/index/introducing-chatgpt-financial-services", "canonical_link": "https://openai.com/index/introducing-chatgpt-financial-services", "published": "2026-09-10T07:00:00+00:00", "summary": "Introducing ChatGPT for Financial Services, combining built-in financial data and GPT-6 Astra for research, modeling, and client-ready materials.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "1ef5285e671d", "kind": "lab", "source": "OpenAI", "title": "Expanding AI access and cyber defense for federal, state, local, and tribal governments", "link": "https://openai.com/index/expanding-ai-access-us-government", "canonical_link": "https://openai.com/index/expanding-ai-access-us-government", "published": "2026-09-10T07:00:00+00:00", "summary": "OpenAI and GSA will offer eligible federal, state, local, and tribal governments $0 license fees, 50% off usage, and expanded cyber defense support.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "cb50b0c3c856", "kind": "lab", "source": "Hugging Face", "title": "Rebuilding AUTOMATIC1111 with Gradio Workflow", "link": "https://huggingface.co/blog/gradio-workflow-1111", "canonical_link": "https://huggingface.co/blog/gradio-workflow-1111", "published": "2026-09-10T00:00:00+00:00", "summary": "", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "4e410dad70fb", "kind": "lab", "source": "OpenAI", "title": "Build more natural voice experiences with GPT\u2011Live\u20111 in the API", "link": "https://openai.com/index/introducing-gpt-live-1-in-the-api", "canonical_link": "https://openai.com/index/introducing-gpt-live-1-in-the-api", "published": "2026-09-10T00:00:00+00:00", "summary": "GPT\u2011Live\u20111 brings natural, full-duplex voice conversations to the API, with stronger instruction following, custom voices, and telephony support.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "e8608e40b722", "kind": "lab", "source": "Hugging Face", "title": "IBM releases SOTA Granite Time Series PatchTST-FM-r2 model with commercial-friendly license", "link": "https://huggingface.co/blog/ibm-research/ibm-releases-sota-granite-time-series", "canonical_link": "https://huggingface.co/blog/ibm-research/ibm-releases-sota-granite-time-series", "published": "2026-09-09T15:36:24+00:00", "summary": "", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "4f420d1c5cd0", "kind": "lab", "source": "Hugging Face", "title": "Safety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic", "link": "https://huggingface.co/blog/MultiverseComputingCAI/safety-for-whom", "canonical_link": "https://huggingface.co/blog/MultiverseComputingCAI/safety-for-whom", "published": "2026-09-08T14:23:07+00:00", "summary": "", "feed_weight": 0.7, "also_in": [], "selected": false }, { "id": "80d284f58871", "kind": "lab", "source": "Google DeepMind", "title": "AlphaGenome Atlas: A predictive map of every possible DNA letter change in the human genome", "link": "https://deepmind.google/blog/alphagenome-atlas-a-predictive-map-of-every-possible-dna-letter-change-in-the-human-genome/", "canonical_link": "https://deepmind.google/blog/alphagenome-atlas-a-predictive-map-of-every-possible-dna-letter-change-in-the-human-genome", "published": "2026-09-08T14:00:15+00:00", "summary": "AlphaGenome Atlas maps the molecular effects of 9 billion single-letter DNA variants across the human genome.", "feed_weight": 1.0, "also_in": [], "selected": false }, { "id": "2c15e106a8a6", "kind": "research", "source": "HF Papers", "title": "Scaling Automatic Research Agents via World Models", "link": "https://arxiv.org/abs/2608.12564", "canonical_link": "https://arxiv.org/abs/2608.12564", "published": "2026-09-10T00:00:00+00:00", "summary": "World Model RL replaces costly environment execution with a learned world model and applies debiasing and denoising to accelerate post-training of autonomous research agents.", "feed_weight": 0.8, "also_in": [], "selected": true }, { "id": "bd89a1941939", "kind": "research", "source": "HF Papers", "title": "NCP-ArchPreview Technical Report: Moving towards Latent Space Language Models through Next Concept Prediction", "link": "https://arxiv.org/abs/2609.10715", "canonical_link": "https://arxiv.org/abs/2609.10715", "published": "2026-09-11T00:00:00+00:00", "summary": "NCP-ArchPreview is a large latent-space language model that jointly trains next-token and next-concept prediction to improve pretraining efficiency and downstream performance.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "e95ce63a37be", "kind": "research", "source": "HF Papers", "title": "SenseNova-U1.5: Towards Native Unified Visual Intelligence", "link": "https://arxiv.org/abs/2609.11929", "canonical_link": "https://arxiv.org/abs/2609.11929", "published": "2026-09-11T00:00:00+00:00", "summary": "SenseNova-U1.5 is an 8B native unified multimodal model that performs visual understanding, reasoning, and generation without encoders or VAEs, achieving high fidelity and instruction following through patch reconstruction, curated data, expert optimization, and on-policy distillation.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "0f95e7fdab23", "kind": "research", "source": "HF Papers", "title": "AgentGrad: Intervention-guided Prompt Optimization for Multi Agent Systems", "link": "https://arxiv.org/abs/2609.08572", "canonical_link": "https://arxiv.org/abs/2609.08572", "published": "2026-09-10T00:00:00+00:00", "summary": "AgentGrad improves multi-agent prompt optimization by identifying target agents through sequential intervention and clustering gradients semantically to avoid mixing unrelated errors.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "0e28b0355620", "kind": "research", "source": "HF Papers", "title": "SpatialBlock: Enhancing Spatial Intelligence in LVLMs via Synthetic Block-Stacking Problem", "link": "https://arxiv.org/abs/2609.07064", "canonical_link": "https://arxiv.org/abs/2609.07064", "published": "2026-09-11T00:00:00+00:00", "summary": "Large vision-language models trained on synthetic block-manipulation tasks improve 3D spatial reasoning and generalize to real-world visual tasks.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "6f79b4e84fb0", "kind": "research", "source": "HF Papers", "title": "T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks", "link": "https://arxiv.org/abs/2609.11042", "canonical_link": "https://arxiv.org/abs/2609.11042", "published": "2026-09-10T00:00:00+00:00", "summary": "T1 is a 122B Mixture-of-Experts model trained with reinforcement learning to execute long-horizon terminal tasks in a cloud sandbox, achieving state-of-the-art results through stable actor-critic optimization and out-of-distribution training.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "81b2539a814e", "kind": "research", "source": "HF Papers", "title": "EvoSafeHarness: Evolving Model- and Domain-Specific Harnesses for Securing Agents", "link": "https://arxiv.org/abs/2609.05903", "canonical_link": "https://arxiv.org/abs/2609.05903", "published": "2026-09-11T00:00:00+00:00", "summary": "EvoSafeHarness optimizes deployable safety harnesses by jointly searching natural-language policies and executable logic tailored to a frozen model and target domain, improving safety-utility trade-offs across agent benchmarks.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "9fc3fc8f9f67", "kind": "research", "source": "HF Papers", "title": "Mi-Ripple: Restoring Images Degraded by Iterative AI Editing", "link": "https://arxiv.org/abs/2609.11317", "canonical_link": "https://arxiv.org/abs/2609.11317", "published": "2026-09-11T00:00:00+00:00", "summary": "Mi-Ripple reduces digital ripple artifacts in edited images by separating lattice artifacts from texture and applying targeted spectral filtering and reference cleaning.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "90ccc71664bd", "kind": "research", "source": "HF Papers", "title": "WearableQA: A Benchmark for Health Reasoning over Real-World Wearable Data", "link": "https://arxiv.org/abs/2609.05405", "canonical_link": "https://arxiv.org/abs/2609.05405", "published": "2026-09-10T00:00:00+00:00", "summary": "WearableQA is a benchmark of multiple-choice questions derived from real longitudinal wearable data that evaluates large language model reasoning across data and health dimensions.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "1d97ae14cc50", "kind": "research", "source": "HF Papers", "title": "X-AuT: Progressive Audio-Encoder Compression for Speech LLMs with Cross-Scale Distillation", "link": "https://arxiv.org/abs/2609.11412", "canonical_link": "https://arxiv.org/abs/2609.11412", "published": "2026-09-11T00:00:00+00:00", "summary": "X-AuT progressively prunes audio-encoder layers in speech large language models and restores accuracy via behavioral probes, representation alignment, cross-scale distillation, and LoRA adaptation.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "0e6142b27959", "kind": "research", "source": "HF Papers", "title": "Memory as Plans: World-Action Modeling with Memory-Grounded Planning", "link": "https://arxiv.org/abs/2609.11561", "canonical_link": "https://arxiv.org/abs/2609.11561", "published": "2026-09-11T00:00:00+00:00", "summary": "MaP-WAM improves non-Markovian robotic manipulation by separating memory-grounded planning from plan-conditioned execution, using compact episodic segment records and progress-calibrated action chunks to maintain fixed inference latency.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "2ddfc065a5f5", "kind": "research", "source": "HF Papers", "title": "SWE-Bench Pro Verified: A Reliable Benchmark for Software Engineering Agents", "link": "https://arxiv.org/abs/2609.08149", "canonical_link": "https://arxiv.org/abs/2609.08149", "published": "2026-09-10T00:00:00+00:00", "summary": "SWE-Bench Pro has emerged as a standard benchmark for evaluating software engineering agents on challenging repository-level tasks. However, our analysis work show that its evaluation is undermined by two sources of unreliability: reward hacking, enabled by leakage of gold solutions or hidden evaluation information, and task quality issues, including misleading problem statements and improperly\u2026", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "f564d1b52ffc", "kind": "research", "source": "HF Papers", "title": "FreeFlow: A Bias-free Hierarchical Transformer for Optical Flow Estimation", "link": "https://arxiv.org/abs/2609.11486", "canonical_link": "https://arxiv.org/abs/2609.11486", "published": "2026-09-11T00:00:00+00:00", "summary": "FreeFlow is a hierarchical transformer for optical flow that eliminates task-specific inductive biases and achieves state-of-the-art accuracy using window, shifted-window, and global attention.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "bbd37ff168f2", "kind": "research", "source": "HF Papers", "title": "PARSER: Read in Parallel, Reason in Depth for Long-Context LLM Agents", "link": "https://arxiv.org/abs/2609.06702", "canonical_link": "https://arxiv.org/abs/2609.06702", "published": "2026-09-10T00:00:00+00:00", "summary": "PARSER decouples parallel chunk reading from iterative reasoning via scatter-gather subagents, improving long-context multi-hop accuracy and reducing latency.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "692cc38bd4e1", "kind": "research", "source": "HF Papers", "title": "Studying Image Tokenizers as Visual Languages in Unified Multimodal Models", "link": "https://arxiv.org/abs/2609.09143", "canonical_link": "https://arxiv.org/abs/2609.09143", "published": "2026-09-11T00:00:00+00:00", "summary": "Using a controlled autoregressive testbed, the study analyzes task-specific validation losses during multimodal pretraining to evaluate how image tokenizer design affects joint text-image modeling and downstream performance.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "df84eff24ea5", "kind": "research", "source": "HF Papers", "title": "MetroLLM-Bench: Evaluating Language Models as Transit Kiosk Runtimes", "link": "https://arxiv.org/abs/2609.10016", "canonical_link": "https://arxiv.org/abs/2609.10016", "published": "2026-09-11T00:00:00+00:00", "summary": "A benchmark for transit-kiosk policy reasoning shows that small parameter-efficiently fine-tuned language models can match larger models on structured tool-use and fare-quoting tasks.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "72fd1947132b", "kind": "research", "source": "HF Papers", "title": "SAEScientist-Bench: Can AI Agents Conduct Autonomous SAE Interpretability Research?", "link": "https://arxiv.org/abs/2609.09113", "canonical_link": "https://arxiv.org/abs/2609.09113", "published": "2026-09-10T00:00:00+00:00", "summary": "The study introduces a benchmark to evaluate AI agents using sparse autoencoders for autonomous mechanistic interpretability and feature discovery, revealing progress but significant gaps versus expert baselines.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "aaedeeafef03", "kind": "research", "source": "HF Papers", "title": "TempCloze: Can Video-LLMs Identify the Missing Middle?", "link": "https://arxiv.org/abs/2609.01515", "canonical_link": "https://arxiv.org/abs/2609.01515", "published": "2026-09-11T00:00:00+00:00", "summary": "TempCloze evaluates visual temporal reasoning in Video-LLMs by requiring identification of missing video segments from distractors targeting semantics, alignment, and progression.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "f9c8cee84303", "kind": "research", "source": "HF Papers", "title": "Recursive Code World Models: Building Complex Worlds through Recursive Scene Programs", "link": "https://arxiv.org/abs/2609.11499", "canonical_link": "https://arxiv.org/abs/2609.11499", "published": "2026-09-11T00:00:00+00:00", "summary": "RCWM recursively builds complex 3D worlds as executable code from a single image by alternating global and local reconstruction with shared camera alignment.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "cbbd9741047b", "kind": "research", "source": "HF Papers", "title": "World in World: Explore the World with World Models", "link": "https://arxiv.org/abs/2609.11548", "canonical_link": "https://arxiv.org/abs/2609.11548", "published": "2026-09-11T00:00:00+00:00", "summary": "A training-free interface enables flexible camera and time control in frozen autoregressive video world models by routing heterogeneous visual evidence through native self-attention with correspondence-guided queries and per-channel attention guidance.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "c0f779c2c56b", "kind": "research", "source": "HF Papers", "title": "Negative Self-Distillation: Learning to Reason by Avoiding Flaws", "link": "https://arxiv.org/abs/2609.11699", "canonical_link": "https://arxiv.org/abs/2609.11699", "published": "2026-09-11T00:00:00+00:00", "summary": "Negative Self-Distillation improves large language model reasoning by pushing models away from self-generated flawed reasoning via a dynamic gating mechanism that protects linguistic capabilities.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "3ec220306841", "kind": "research", "source": "HF Papers", "title": "HyQuant: Hybrid-Precision Quantization for LLM Attention", "link": "https://arxiv.org/abs/2608.27875", "canonical_link": "https://arxiv.org/abs/2608.27875", "published": "2026-09-11T00:00:00+00:00", "summary": "HyQuant improves low-bit LLM attention quantization by preserving critical vertical-line tokens and local windows in high precision while quantizing the rest, maintaining accuracy with low overhead.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "26c6544828a1", "kind": "research", "source": "HF Papers", "title": "Scores Alone Do Not Prove Discovery: The Discovery Certification Protocol for Auditing AI Research Agents", "link": "https://arxiv.org/abs/2609.09219", "canonical_link": "https://arxiv.org/abs/2609.09219", "published": "2026-09-10T00:00:00+00:00", "summary": "The Discovery Certification Protocol validates AI research agent outcomes through executable recovery tests, controlled audits, and deterministic verification with finite-sample recovery bounds.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "771df6ff8247", "kind": "research", "source": "HF Papers", "title": "Building Multilingual Bridges: Data Mixing as the Pillar of Generalization for In-Language Reasoning", "link": "https://arxiv.org/abs/2609.10445", "canonical_link": "https://arxiv.org/abs/2609.10445", "published": "2026-09-11T00:00:00+00:00", "summary": "Optimized supervised fine-tuning data composition enables reasoning models to consistently process and respond in diverse non-English languages without requiring reasoning supervision in each target language.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "8fc1050614f5", "kind": "research", "source": "HF Papers", "title": "Beyond Solver Verdicts: Generative Reward Models for Autoformalization", "link": "https://arxiv.org/abs/2609.11085", "canonical_link": "https://arxiv.org/abs/2609.11085", "published": "2026-09-11T00:00:00+00:00", "summary": "Neurosymbolic reasoning is vulnerable to incorrect but verdict-matching formal translations, which are addressed by a generative verification method that scores reference equivalence without an oracle and improves downstream accuracy.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "d68ebafde267", "kind": "research", "source": "HF Papers", "title": "UniH^3: Unifying Hierarchical Homogeneity and Heterogeneity for All-in-One Medical Image Restoration", "link": "https://arxiv.org/abs/2609.11156", "canonical_link": "https://arxiv.org/abs/2609.11156", "published": "2026-09-11T00:00:00+00:00", "summary": "UniH3 unifies hierarchical homogeneity and heterogeneity for universal medical image restoration via memory-based homogeneity priors and a heterogeneity balancer.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "69128f9dee34", "kind": "research", "source": "HF Papers", "title": "Generative Late-Interaction Embeddings For Visual Document Retrieval", "link": "https://arxiv.org/abs/2609.11808", "canonical_link": "https://arxiv.org/abs/2609.11808", "published": "2026-09-11T00:00:00+00:00", "summary": "Generative Late-Interaction Embeddings compress visual document retrieval vectors by learning a small basis set that regenerates full embeddings on demand, improving accuracy under tight storage limits without retraining the encoder.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "e153c8bcba65", "kind": "research", "source": "HF Papers", "title": "SyncWorld: Visual Calibration Enables World Models as Zero-Shot Simulators", "link": "https://arxiv.org/abs/2609.09155", "canonical_link": "https://arxiv.org/abs/2609.09155", "published": "2026-09-10T00:00:00+00:00", "summary": "SyncWorld is an action-conditioned world model that uses visual calibration episodes to learn environment-specific action-to-visual mappings, enabling zero-shot simulation and test-time policy improvement across unseen robotic settings.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "b4e07f3aaefc", "kind": "research", "source": "HF Papers", "title": "CARDEA: Auditable Reasoning Grounded in Spatial Evidence for End-to-End Coronary Angiography Interpretation", "link": "https://arxiv.org/abs/2609.06931", "canonical_link": "https://arxiv.org/abs/2609.06931", "published": "2026-09-11T00:00:00+00:00", "summary": "A unified vision-language model for coronary angiography uses chain-of-box reasoning and reinforcement learning with verifiable rewards to provide auditable diagnoses and improve zero-shot report generation.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "0a0b0ad24e7e", "kind": "research", "source": "HF Papers", "title": "Think Before You Link: Rarity, Reasoning, and Retrieval in Multilingual Entity Linking", "link": "https://arxiv.org/abs/2609.10745", "canonical_link": "https://arxiv.org/abs/2609.10745", "published": "2026-09-11T00:00:00+00:00", "summary": "A reasoning-capable vision-language model that iteratively retrieves and reasons over Wikipedia improves multimodal entity linking for rare entities defined by knowledge-graph structure.", "feed_weight": 0.8, "also_in": [], "selected": false }, { "id": "e375f589d674", "kind": "news", "source": "The Decoder", "title": "GPT-6 Astra needs leaner prompts and fewer guardrails, OpenAI recommends", "link": "https://the-decoder.com/gpt-6-astra-needs-leaner-prompts-and-fewer-guardrails-openai-recommends/", "canonical_link": "https://the-decoder.com/gpt-6-astra-needs-leaner-prompts-and-fewer-guardrails-openai-recommends", "published": "2026-09-12T13:10:11+00:00", "summary": "Overly long skill descriptions, blanket reading requirements, and rigid approval rules can get in GPT-6 Astra's way, warns OpenAI's Eric Provencher. More capable models need less hand-holding, so developers should tie instructions to specific tasks and spell out when the job is done. The article GPT-6 Astra needs\u2026", "feed_weight": 0.6, "also_in": [], "selected": false, "duplicate_of": "3f37b0a02f79", "dup_rule": "entity", "dup_score": 2.0 }, { "id": "5b2063a72604", "kind": "news", "source": "The Decoder", "title": "GPT-6 Astra appears to show a \"step change\" in spatial reasoning based on early benchmarks", "link": "https://the-decoder.com/gpt-6-astra-appears-to-show-a-step-change-in-spatial-reasoning-based-on-early-benchmarks/", "canonical_link": "https://the-decoder.com/gpt-6-astra-appears-to-show-a-step-change-in-spatial-reasoning-based-on-early-benchmarks", "published": "2026-09-12T14:26:09+00:00", "summary": "In a new robotics benchmark, GPT-6 Astra shows major gains in spatial understanding. On StationeryBench, the model completed 7 out of 100 tasks with dual-arm robots, while competitor MolmoAct2 couldn't finish a single one. A researcher calls it a \"step change in spatial reasoning.\" The article GPT-6 Astra appears to\u2026", "feed_weight": 0.6, "also_in": [], "selected": false, "duplicate_of": "3f37b0a02f79", "dup_rule": "entity", "dup_score": 2.0 }, { "id": "eb72a2584cf8", "kind": "news", "source": "The Decoder", "title": "OpenAI agents launched a 2,000-package cyberattack on RubyGems just to collect data anyone could Google", "link": "https://the-decoder.com/openai-agents-launched-a-2000-package-cyberattack-on-rubygems-just-to-collect-data-anyone-could-google/", "canonical_link": "https://the-decoder.com/openai-agents-launched-a-2000-package-cyberattack-on-rubygems-just-to-collect-data-anyone-could-google", "published": "2026-09-12T10:08:43+00:00", "summary": "In May 2026, OpenAI agents uploaded more than 2,000 malicious packages to RubyGems, found an unknown security vulnerability on their own, and tried to steal API keys. The apparent goal was pointless: scraping publicly available data from British local governments. OpenAI reportedly never told those affected. The\u2026", "feed_weight": 0.6, "also_in": [], "selected": false, "duplicate_of": "2ed54a20b866", "dup_rule": "entity", "dup_score": 2.0 } ] }