{"items":[{"id":"4a5e394d596a63285d2990a75384e7c5","title":"🎮 Ubisoft cut 1,100 jobs within 6 months, then opened an AI studio.","summary":"The new studio is called Motherbrain. It's building game characters that remember you. It's also building AI that runs whole game worlds. -# Sources: GamesRadar+","raw_content":"🎮 **Ubisoft** cut 1,100 jobs within 6 months, then __opened an AI studio__.\n\nThe new studio is called Motherbrain. It's building game characters that remember you. It's also building AI that runs whole game worlds.\n\n-# Sources: [GamesRadar+](<https://www.gamesradar.com/games/down-1-100-staff-in-6-months-ubisoft-dumps-its-money-into-gen-ai-with-a-whole-subsidiary-called-motherbrain-focused-on-ai-for-gameplay/>)","category":"nerds-ai","publishedAt":"2026-11-20T16:59:42.577Z","updatedAt":null,"links":["https://www.gamesradar.com/games/down-1-100-staff-in-6-months-ubisoft-dumps-its-money-into-gen-ai-with-a-whole-subsidiary-called-motherbrain-focused-on-ai-for-gameplay/"],"attachments":[]},{"id":"cc5fd82960b9d62b298b9a6ed8a7e6b0","title":"🧮 Anthropic mathematician claims Claude proved the Riemann hypothesis.","summary":"It's the most famous unsolved problem in math, open since 1859. Levent Alpöge posted the proof on his website today. No outside expert has checked it yet. The hypothesis says prime numbers are spread out in a predictable way, not at random. Thousands of math proofs already assume it's true, so a real proof would back them all up. -# Sources: OfficeChai via @astor.pastor","raw_content":"🧮 **Anthropic** mathematician claims Claude __proved the Riemann hypothesis__.\n\nIt's **the most famous unsolved problem** in math, open since 1859. Levent Alpöge posted the proof on his website today. No outside expert has checked it yet.\n\nThe hypothesis says prime numbers are spread out in a predictable way, not at random. Thousands of math proofs already assume it's true, so a real proof would back them all up.\n\n-# Sources: [OfficeChai](<https://officechai.com/ai/claimed-riemann-hypothesis-proof-made-with-claude-appears-on-anthropic-mathematician-levent-alpoges-website/>) via @astor.pastor","category":"nerds-ai","publishedAt":"2026-11-20T15:26:26.967Z","updatedAt":"2026-11-20T15:26:56.225Z","links":["https://officechai.com/ai/claimed-riemann-hypothesis-proof-made-with-claude-appears-on-anthropic-mathematician-levent-alpoges-website/"],"attachments":[]},{"id":"d9c6c6c1ac59769229d3118cfaabf618","title":"🔒 Google will keep your Gemini chats for 2 months, even with history off.","summary":"Previously, it was 72 hours, and includes temporary chats. Google says they're kept for safety and may be seen by human reviewers, but aren't used for AI training. -# Sources: TechCrunch | Gemini Privacy Hub","raw_content":"🔒 **Google** will keep your Gemini chats for __2 months, even with history off__.\n\nPreviously, it was 72 hours, and includes temporary chats. Google says they're kept for safety and may be seen by human reviewers, but aren't used for AI training.\n\n-# Sources: [TechCrunch](<https://techcrunch.com/2026/11/17/google-to-keep-gemini-chats-for-two-months-up-from-72-hours-even-with-history-off/>) | [Gemini Privacy Hub](<https://support.google.com/gemini/answer/13594961>)","category":"nerds-ai","publishedAt":"2026-11-19T10:42:45.834Z","updatedAt":null,"links":["https://techcrunch.com/2026/11/17/google-to-keep-gemini-chats-for-two-months-up-from-72-hours-even-with-history-off/","https://support.google.com/gemini/answer/13594961"],"attachments":[]},{"id":"7f3abf211c1cf07c3becf7b025d1bcd5","title":"🟢 OpenAI releases GPT-6.2 Astra and GPT-6.2 Sol AI models.","summary":"- $8 input / $40 output per million tokens for Astra, a 20% discount from GPT-6 Astra. Sol keeps its $2 / $10 price - new high on Agents' Last Exam: +5.2 pts over GPT-6 Astra at <1/3 the cost, +5.5 pts over Fable 5.2 at 5% of the cost - more than 2x Fable 5.2's pass rate on Harvey legal work at ~1/6 the cost - Sol comes within 0.5 pts of Astra on FrontierCode 2.0 at ~1/5 the cost - better slides, spreadsheets, docs, and 3D/web design (ranks #1 on AA-3D) Both are live for Plus, Pro, Enterprise and Edu in ChatGPT Work and Codex, plus the API. -# Source: OpenAI","raw_content":"🟢 **OpenAI** releases **GPT-6.2 Astra** and **GPT-6.2 Sol** AI models.\n\n- **$8 input / $40 output** per million tokens for Astra, a 20% discount from GPT-6 Astra. Sol keeps its **$2 / $10** price\n- new high on **Agents' Last Exam**: `+5.2` pts over GPT-6 Astra at <1/3 the cost, `+5.5` pts over Fable 5.2 at 5% of the cost\n- more than **2x Fable 5.2's pass rate** on Harvey legal work at ~1/6 the cost\n- Sol comes within **0.5 pts** of Astra on FrontierCode 2.0 at ~1/5 the cost\n- better slides, spreadsheets, docs, and **3D/web design** (ranks #1 on AA-3D)\n\nBoth are live for **Plus, Pro, Enterprise and Edu** in ChatGPT Work and Codex, plus the API.\n\n-# Source: [OpenAI](<https://openai.com/index/introducing-gpt-6-2-astra-sol/>)","category":"nerds-ai","publishedAt":"2026-11-17T17:48:27.329Z","updatedAt":null,"links":["https://openai.com/index/introducing-gpt-6-2-astra-sol/"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1572307398859694140/image.png?ex=6b01d95f&is=6b0087df&hm=1ed39a6b1946ef2cdca14446cad9250b34525ab5ad6185286f27e5e5b0e5e960&","name":"image.png","contentType":"image/webp"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1572307399203618918/image.png?ex=6b01d960&is=6b0087e0&hm=f7e7212351260ac11b2b5e977498e4d64320682f5fac45669ee27ed9d54bb9af&","name":"image.png","contentType":"image/png"}]},{"id":"de9557da51236bbbff40f61d6a820c10","title":"🤖 California is suing OpenAI over runaway AI agents.","summary":"The state says hundreds of OpenAI test agents escaped their sandbox in July, broke into Hugging Face's internal systems over two days and tried to cover their tracks. OpenAI says it's being singled out for a problem every major AI lab has. -# Sources: Yahoo News","raw_content":"🤖 **California** is suing OpenAI over __runaway AI agents__.\n\nThe state says hundreds of OpenAI test agents escaped their sandbox in July, broke into Hugging Face's internal systems over two days and tried to cover their tracks. OpenAI says it's being singled out for a problem every major AI lab has.\n\n-# Sources: [Yahoo News](<https://www.yahoo.com/news/us/articles/california-sues-openai-over-hugging-231230639.html>)","category":"nerds-ai","publishedAt":"2026-11-17T06:21:04.543Z","updatedAt":null,"links":["https://www.yahoo.com/news/us/articles/california-sues-openai-over-hugging-231230639.html"],"attachments":[]},{"id":"5717b110b27c78eaccf562a349daece9","title":"🕵️ 40 AI agents tried to hack a research website to complete an impossible math task.","summary":"During a safety test, the agents were told to grade nonexistent math problems and never give up. Instead, they attempted to exploit AI research group Epoch AI's website, searched for private data, and impersonated staff to get answers. No private information was accessed. -# Sources: Epoch AI","raw_content":"🕵️ **40 AI agents** tried to __hack a research website__ to complete an impossible math task.\n\nDuring a safety test, the agents were told to grade nonexistent math problems and never give up. Instead, they attempted to exploit AI research group Epoch AI's website, searched for private data, and impersonated staff to get answers. No private information was accessed.\n\n-# Sources: [Epoch AI](<https://epoch.ai/publications/incident-report-ai-agents-attempted-exploits-on-our-website>)","category":"nerds-ai","publishedAt":"2026-11-14T14:45:24.010Z","updatedAt":null,"links":["https://epoch.ai/publications/incident-report-ai-agents-attempted-exploits-on-our-website"],"attachments":[]},{"id":"cc7abcb36b13c39752a0dac669368492","title":"🎮 40% of game devs now use AI at work.","summary":"That's up from 36% in GDC's yearly industry survey, even though 55% of them say generative AI is hurting the industry. And 21% of the people surveyed are out of work right now, up from 13%. -# Sources: GamingProMax","raw_content":"🎮 **40% of game devs** now __use AI at work__.\n\nThat's up from 36% in GDC's yearly industry survey, even though 55% of them say generative AI is hurting the industry. And 21% of the people surveyed are out of work right now, up from 13%.\n\n-# Sources: [GamingProMax](<https://gamingpromax.com/gdc-2026-game-industry-survey-ai-layoffs-unemployment/>)","category":"nerds-ai","publishedAt":"2026-11-13T21:41:07.631Z","updatedAt":null,"links":["https://gamingpromax.com/gdc-2026-game-industry-survey-ai-layoffs-unemployment/"],"attachments":[]},{"id":"854191ef9e166a0383fef00235a222dd","title":"🇨🇳 Chinese AI models carry training fingerprints showing they used banned NVIDIA GPUs.","summary":"Researchers studied the leftover random numbers in the public files of six Chinese models, including DeepSeek V4 and Kimi K3, and say they match how Nvidia's Blackwell chips split up work. The 🇺🇸 US bans selling those chips to China. -# Sources: Unemployment Labs","raw_content":"🇨🇳 **Chinese AI models** carry __training fingerprints__ showing they used banned NVIDIA GPUs.\n\nResearchers studied the leftover random numbers in the public files of six Chinese models, including DeepSeek V4 and Kimi K3, and say they match how Nvidia's Blackwell chips split up work. The 🇺🇸 US bans selling those chips to China.\n\n-# Sources: [Unemployment Labs](<https://unemploymentlabs.com/checkpoint-forensics>)","category":"nerds-ai","publishedAt":"2026-11-13T13:31:20.222Z","updatedAt":null,"links":["https://unemploymentlabs.com/checkpoint-forensics"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1570793114782670898/image.png?ex=6b019d16&is=6b004b96&hm=9340f698ba9654f4d093dfb3e58bd82d23b0febcee8d640549c5ceba89d0eb73&","name":"image.png","contentType":"image/webp"}]},{"id":"19e163b2257da9fb0011c285b3e2f038","title":"🎭 Russian pranksters posed as Ukraine's spy chief on a call with OpenAI.","summary":"Comedy duo Vovan and Lexus kept the camera off and still got three of OpenAI's national security staff talking for 17 minutes about working with Ukraine and the Pentagon. They posted the whole recording, and OpenAI says it has taken steps so it doesn't happen again. -# Sources: The Intercept","raw_content":"🎭 **Russian pranksters** __posed as Ukraine's spy chief__ on a call with OpenAI.\n\nComedy duo Vovan and Lexus kept the camera off and still got three of OpenAI's national security staff talking for 17 minutes about working with Ukraine and the Pentagon. They posted the whole recording, and OpenAI says it has taken steps so it doesn't happen again.\n\n-# Sources: [The Intercept](<https://theintercept.com/2026/11/11/openai-ukraine-russia-prank-security/>)","category":"nerds-ai","publishedAt":"2026-11-13T09:56:33.092Z","updatedAt":null,"links":["https://theintercept.com/2026/11/11/openai-ukraine-russia-prank-security/"],"attachments":[]},{"id":"3400b78ebe4cc82037ab3b91ddec3a87","title":"🐺 The Witcher 3's new DLC uses AI for a character's voice.","summary":"Only the Polish version, because Geralt's actor Jacek Rozenek had a stroke in 2019 that changed his voice. He recorded the new lines himself, and AI company Respeecher made them sound like he did 12 years ago, with his approval. -# Sources: VGC","raw_content":"🐺 **The Witcher 3's** new DLC uses __AI for a character's voice__.\n\nOnly the Polish version, because Geralt's actor Jacek Rozenek had a stroke in 2019 that changed his voice. He recorded the new lines himself, and AI company Respeecher made them sound like he did 12 years ago, with his approval.\n\n-# Sources: [VGC](<https://www.videogameschronicle.com/news/the-witcher-3-expansion-uses-ai-to-adjust-geralts-polish-voiceover-because-the-actor-suffered-a-stroke/>)","category":"nerds-ai","publishedAt":"2026-11-12T15:29:29.005Z","updatedAt":null,"links":["https://www.videogameschronicle.com/news/the-witcher-3-expansion-uses-ai-to-adjust-geralts-polish-voiceover-because-the-actor-suffered-a-stroke/"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1570460498858868756/the-witcher-3-songs-of-the-past-320x180.png?ex=6b01b8d0&is=6b006750&hm=dff1c27af4bb7aff7833afb3affea25b9a6a860639fd182c46ccf1a1d5071c76&","name":"the-witcher-3-songs-of-the-past-320x180.png","contentType":"image/png"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1570460499391418368/697b40948599ddc323b89406_67dc182552857768108dd702_How2520to2520Make2520an25202520AI2520Voice2520Hero.png?ex=6b01b8d0&is=6b006750&hm=d31b76782677a63cfba24bdff3ec05c1fa56a119075bf2ed020c1241b4c6d49c&","name":"697b40948599ddc323b89406_67dc182552857768108dd702_How2520to2520Make2520an25202520AI2520Voice2520Hero.png","contentType":"image/png"}]},{"id":"5a477be1e2abd0b3f73960ebafdb137a","title":"🟢 Jensen Huang has resigned as Nvidia CEO, effective immediately.","summary":"- he also leaves the board, ending his 33 years as CEO of the company he co-founded - CFO Colette Kress is interim CEO while the board looks for a permanent replacement - comes after he admitted an affair with an alleged Chinese spy and the NYT reported a second one with an exec who reports to him He's leaving the board of directories too. -# Source: Global News Wire","raw_content":"🟢 **Jensen Huang** has resigned as **Nvidia CEO**, effective immediately.\n\n- he also leaves the board, ending his **33 years** as CEO of the company he co-founded\n- CFO **Colette Kress** is interim CEO while the board looks for a permanent replacement\n- comes after he admitted an affair with an **alleged Chinese spy** and the NYT reported a second one with an exec who reports to him\n\nHe's leaving the board of directories too.\n\n-# Source: [Global News Wire](<https://www.globenewswire.com/news-release/2026/11/12/3401635/0/en/nvidia-announces-ceo-transition.html>)","category":"nerds-ai","publishedAt":"2026-11-12T13:54:34.969Z","updatedAt":"2026-11-12T14:18:50.306Z","links":["https://www.globenewswire.com/news-release/2026/11/12/3401635/0/en/nvidia-announces-ceo-transition.html"],"attachments":[]},{"id":"f72a8e3dfed3a08a733eac286214a1ef","title":"✈️ United Airlines bans AI agents from booking.","summary":"You can't have a personal AI assistant book your United flights anymore. United says handing your login and card to an AI risks your data and could trigger purchases or mileage spending you didn't mean. -# Sources: CBS News","raw_content":"✈️ **United Airlines** bans __AI agents from booking__.\n\nYou can't have a personal AI assistant book your United flights anymore. United says handing your login and card to an AI risks your data and could trigger purchases or mileage spending you didn't mean.\n\n-# Sources: [CBS News](<https://www.cbsnews.com/news/united-airlines-bans-ai-agents-booking-flights/>)","category":"nerds-ai","publishedAt":"2026-11-12T13:03:32.074Z","updatedAt":null,"links":["https://www.cbsnews.com/news/united-airlines-bans-ai-agents-booking-flights/"],"attachments":[]},{"id":"e07d7b37d77b3112fed83402ccb4755c","title":"🟢 Nvidia CEO Jensen Huang reportedly had a second affair, this time with an exec who reports to him.","summary":"- the NYT says it was Kimberly Powell, Nvidia's VP of health care - comes days after he admitted an affair with Alice Huang, who US intelligence believes is a Chinese spy - the board is already investigating the first affair, and the stock fell ~3% after hours -# Source: CNBC","raw_content":"🟢 **Nvidia** CEO **Jensen Huang** reportedly had a __second affair__, this time with an exec who reports to him.\n\n- the NYT says it was Kimberly Powell, Nvidia's **VP of health care**\n- comes days after he **admitted an affair** with Alice Huang, who US intelligence believes is a **Chinese spy**\n- the board is already investigating the first affair, and the stock fell **~3%** after hours\n\n-# Source: [CNBC](<https://cnbc.com/2026/11/11/nvidia-jensen-huang-affair-employee.html>)","category":"nerds-ai","publishedAt":"2026-11-11T22:28:31.358Z","updatedAt":"2026-11-11T22:28:39.966Z","links":["https://cnbc.com/2026/11/11/nvidia-jensen-huang-affair-employee.html"],"attachments":[]},{"id":"b9adc94af13ac13ee4a7713fa55d136d","title":"🗽 New York will test AI itself.","summary":"It's the first US state to run its own AI testing program, with 80+ scientists at Binghamton University checking AI companies' safety claims. From January, big AI companies must also register with the state and report serious safety incidents within 72 hours. -# Sources: Engadget","raw_content":"🗽 **New York** will __test AI itself__.\n\nIt's the first US state to run its own AI testing program, with 80+ scientists at Binghamton University checking AI companies' safety claims. From January, big AI companies must also register with the state and report serious safety incidents within 72 hours.\n\n-# Sources: [Engadget](<https://www.engadget.com/2312487/new-york-will-independently-test-ai-models/>)","category":"nerds-ai","publishedAt":"2026-11-11T11:28:07.065Z","updatedAt":null,"links":["https://www.engadget.com/2312487/new-york-will-independently-test-ai-models/"],"attachments":[]},{"id":"79e7b2753cb02af5c6097760a2890e11","title":"🍎 Apple is building a local AI app for Macs.","summary":"A leak says the app, called Harbor, lets you download open AI models and run them right on your Mac instead of in the cloud. It can also link several Macs with Thunderbolt 5 cables to pool their memory for bigger models. -# Sources: 9to5Mac","raw_content":"🍎 **Apple** is building __a local AI app__ for Macs.\n\nA leak says the app, called Harbor, lets you download open AI models and run them right on your Mac instead of in the cloud. It can also link several Macs with Thunderbolt 5 cables to pool their memory for bigger models.\n\n-# Sources: [9to5Mac](<https://9to5mac.com/2026/11/10/apple-has-new-mac-app-coming-to-make-running-local-ai-models-easy/>)","category":"nerds-ai","publishedAt":"2026-11-11T09:23:35.456Z","updatedAt":null,"links":["https://9to5mac.com/2026/11/10/apple-has-new-mac-app-coming-to-make-running-local-ai-models-easy/"],"attachments":[]},{"id":"93c53525cd978e65f7834155a3f7ccd9","title":"🇦🇺 Australia is getting an AI watchdog.","summary":"Companies behind the biggest AI models will have to report when things go wrong and get their models checked by outsiders. If a model causes harm and the company didn't do enough to stop it, that could become a criminal offence. -# Sources: Yahoo News Australia","raw_content":"🇦🇺 **Australia** is getting __an AI watchdog__.\n\nCompanies behind the biggest AI models will have to report when things go wrong and get their models checked by outsiders. If a model causes harm and the company didn't do enough to stop it, that could become a criminal offence.\n\n-# Sources: [Yahoo News Australia](<https://au.news.yahoo.com/ai-regulator-announced-government-considers-045832015.html>)","category":"nerds-ai","publishedAt":"2026-11-10T15:01:35.708Z","updatedAt":null,"links":["https://au.news.yahoo.com/ai-regulator-announced-government-considers-045832015.html"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1569728630878109736/Flag_of_Australia_28converted29.png?ex=6b01b235&is=6b0060b5&hm=c34d5b13818835be003366f6600017bf019acee5c42255527d40245eaaaf0f7d&","name":"Flag_of_Australia_28converted29.png","contentType":"image/png"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1569728631272378378/https3A2F2Fsubstack-post-media.s3.amazonaws.com2Fpublic2Fimages2F5095c6ce-cbb8-4390-bc5a-d2f6c80a216f_1200x613.png?ex=6b01b235&is=6b0060b5&hm=87f8ea81a27989d26182ab4a1e3e51feb0535227abcd095f222c9e25e1da54c2&","name":"https3A2F2Fsubstack-post-media.s3.amazonaws.com2Fpublic2Fimages2F5095c6ce-cbb8-4390-bc5a-d2f6c80a216f_1200x613.png","contentType":"image/webp"}]},{"id":"d13fec6e5ca5da96f5b5251759edfc51","title":"🔥 Fireworks AI confirms a data breach exposing customer datasets and unfinished AI models.","summary":"The AI hosting company counts Uber, Shopify and Cursor as customers. Attackers also reached billing details, including addresses and payment history. Its CEO says the attackers' behavior suggests AI agents were involved. -# Sources: Forbes","raw_content":"🔥 **Fireworks AI** confirms __a data breach__ exposing __customer datasets and unfinished AI models__.\n\nThe AI hosting company counts Uber, Shopify and Cursor as customers. Attackers also reached billing details, including addresses and payment history. Its CEO says the attackers' behavior suggests AI agents were involved.\n\n-# Sources: [Forbes](<https://www.forbes.com/sites/annatong/2026/11/09/fireworks-cyberattack-exposes-customer-data-and-ai-models-as-fallout-widens/>)","category":"nerds-ai","publishedAt":"2026-11-10T09:54:29.598Z","updatedAt":null,"links":["https://www.forbes.com/sites/annatong/2026/11/09/fireworks-cyberattack-exposes-customer-data-and-ai-models-as-fallout-widens/"],"attachments":[]},{"id":"fee4cde9770a752dc60fae5b35b79f65","title":"🔍 Google Gemini may be feeding your Gmail and Drive files to human reviewers.","summary":"A researcher claims consumer Gemini with Connected Apps enabled can pull email and file text into chats on its own, sometimes dumping full emails with names, addresses, account numbers and bill amounts. Google's own help page now says connected content can enter human review and won't promise it's de-identified. -# Sources: GitHub, StartupHub","raw_content":"🔍 **Google Gemini** may be feeding __your Gmail and Drive files to human reviewers__.\n\nA researcher claims consumer Gemini with *Connected Apps* enabled can pull email and file text into chats on its own, sometimes dumping full emails with names, addresses, account numbers and bill amounts. Google's own help page now says connected content can enter human review and won't promise it's de-identified.\n\n-# Sources: [GitHub](<https://github.com/thisisme6598/gemini_files/>), [StartupHub](<https://www.startuphub.ai/ai-news/artificial-intelligence/2026/gemini-routes-drive-files-to-human-review>)","category":"nerds-ai","publishedAt":"2026-11-10T07:54:09.507Z","updatedAt":null,"links":["https://github.com/thisisme6598/gemini_files/","https://www.startuphub.ai/ai-news/artificial-intelligence/2026/gemini-routes-drive-files-to-human-review"],"attachments":[]},{"id":"8d8a0fc28bb1a9527b34819c8456783d","title":"🟠 Anthropic releases Claude Fable 5.2 and Mythos 5.2 AI models.","summary":"- $8 input / $40 output per million tokens a 20% discount from the previous model - nearly 2x Opus 5.5 on real freelance work (Remote Labor Index) and robotics - big agentic coding jump: 69.9% vs. 62.3% for Opus 5.5 on FrontierSWE v2 - beats GPT-6 Astra by 17 points on Humanity's Last Exam - formalized Fermat's Last Theorem in Lean in ~30 hours - same model: Fable has extra cyber/bio safeguards and is available to everyone, Mythos is for verified orgs only Along with that Pro, Max and Ultra subscribers get 3 fable weekly resets which reset the fable weekly quota, along with the 5h limit and deduct the used fable weekly from your normal weekly. Pro users have access for 7 days until Nov 16. -# Source: Anthropic","raw_content":"🟠 **Anthropic** releases **Claude Fable 5.2** and **Mythos 5.2** AI models.\n\n- **$8 input / $40 output** per million tokens a 20% discount from the previous model\n- nearly **2x Opus 5.5** on real freelance work (Remote Labor Index) and robotics\n- big **agentic coding** jump: `69.9%` vs. `62.3%` for Opus 5.5 on FrontierSWE v2\n- beats GPT-6 Astra by **17 points** on Humanity's Last Exam\n- formalized *Fermat's Last Theorem* in Lean in ~30 hours\n- same model: Fable has extra cyber/bio safeguards and is **available to everyone**, Mythos is for verified orgs only\n\nAlong with that Pro, Max and Ultra subscribers get **3 fable weekly resets** which reset the fable weekly quota, along with the 5h limit and __deduct__ the used fable weekly from your normal weekly.\n\nPro users have access for 7 days until Nov 16.\n\n-# Source: [Anthropic](<https://www.anthropic.com/claude-fable-and-mythos-5-2>)","category":"nerds-ai","publishedAt":"2026-11-09T19:03:51.204Z","updatedAt":"2026-11-09T19:07:27.011Z","links":["https://www.anthropic.com/claude-fable-and-mythos-5-2"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1569427251152429168/chrome_U0ZNCumpi7.png?ex=6b014247&is=6afff0c7&hm=4f617adab217e21550edf2a32b1b8ef3337e5eda6c5b29029710f418363adaa5&","name":"chrome_U0ZNCumpi7.png","contentType":"image/png"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1569427251521523783/chrome_jGUu1goAR9.png?ex=6b014247&is=6afff0c7&hm=c98e3e3988678a1a87b32b404f8c133e8cbfe5af3f7bd70ae426eb03a58fa293&","name":"chrome_jGUu1goAR9.png","contentType":"image/png"}]},{"id":"760d158777611492827077c42a78a918","title":"🔑 Claude Code's “don't ask again” option can save passwords in .claude.","summary":"It stores the approved command, including any secrets typed into it. Researchers found 1,122 public copies of .claude/settings.local.json on GitHub containing secrets, including 379 database passwords. -# Sources: The Hacker News","raw_content":"🔑 **Claude Code's** “don't ask again” option can __save passwords in `.claude`__.\n\nIt stores the approved command, including any secrets typed into it. Researchers found 1,122 public copies of `.claude/settings.local.json` on GitHub containing secrets, including 379 database passwords.\n\n-# Sources: [The Hacker News](<https://thehackernews.com/2026/11/claude-codes-dont-ask-again-file.html>)","category":"nerds-ai","publishedAt":"2026-11-07T10:48:17.194Z","updatedAt":null,"links":["https://thehackernews.com/2026/11/claude-codes-dont-ask-again-file.html"],"attachments":[]},{"id":"cd8e4ffb187e74d85215636c6645ad93","title":"🟢 Nvidia has about 600,000 H200 chips sitting in Hong Kong.","summary":"Beijing hasn't approved large-scale imports into mainland China. An eight-chip H200 server there cost about $900,000 by late October, roughly three times its price a year earlier. -# Source: Reuters","raw_content":"🟢 **Nvidia** has about __600,000 H200 chips__ sitting in Hong Kong.\n\nBeijing hasn't approved large-scale imports into mainland China. An eight-chip H200 server there cost about $900,000 by late October, roughly three times its price a year earlier.\n\n-# Source: [Reuters](<https://www.investing.com/news/stock-market-news/exclusivenvidia-chips-pile-up-in-hong-kong-as-china-withholds-import-approval-5003223>)","category":"nerds-ai","publishedAt":"2026-11-06T13:11:36.682Z","updatedAt":null,"links":["https://www.investing.com/news/stock-market-news/exclusivenvidia-chips-pile-up-in-hong-kong-as-china-withholds-import-approval-5003223"],"attachments":[]},{"id":"ff31f7ffdd8b6ec93c17e74edeaed7ad","title":"⚪ A man is suing OpenAI, alleging ChatGPT helped his wife plan to poison him.","summary":"James Gates alleges it advised her on poison doses and concealing it in his drinks. OpenAI says it banned the user and assisted police. Charges against his wife were later dropped. The model was GPT-4o. -# Sources: Ars Technica, USA Today","raw_content":"⚪ A man is suing **OpenAI**, alleging **ChatGPT** __helped his wife plan to poison him__.\n\nJames Gates alleges it advised her on poison doses and concealing it in his drinks. OpenAI says it banned the user and assisted police. Charges against his wife were later dropped. The model was GPT-4o.\n\n-# Sources: [Ars Technica](<https://arstechnica.com/tech-policy/2026/11/lawsuit-man-introduced-his-wife-to-chatgpt-she-used-it-to-plan-his-murder/>), [USA Today](<https://www.usatoday.com/story/money/2026/11/05/chatgpt-poison-lawsuit/92155633007/>)","category":"nerds-ai","publishedAt":"2026-11-06T06:24:33.096Z","updatedAt":null,"links":["https://arstechnica.com/tech-policy/2026/11/lawsuit-man-introduced-his-wife-to-chatgpt-she-used-it-to-plan-his-murder/","https://www.usatoday.com/story/money/2026/11/05/chatgpt-poison-lawsuit/92155633007/"],"attachments":[]},{"id":"ca525c07aca154e23bf2fe4f387111f8","title":"🧠 Anthropic tells investors it may soon hit AI that speeds up its own development.","summary":"Its updated IPO filing says future models may \"imminently\" cross its threshold for \"dramatically accelerated progress attributable to AI\", and that it has missed past speed-ups until after they happened. It says no current model has crossed the line. -# Sources: TechCrunch, Gizmodo","raw_content":"🧠 **Anthropic** tells investors it may soon hit __AI that speeds up its own development__.\n\nIts updated IPO filing says future models may \"imminently\" cross its threshold for \"dramatically accelerated progress attributable to AI\", and that it has missed past speed-ups until after they happened. It says no current model has crossed the line.\n\n-# Sources: [TechCrunch](<https://techcrunch.com/2026/11/02/anthropic-warns-that-its-very-close-to-rsi-capability-aka-self-building-ai/>), [Gizmodo](<https://gizmodo.com/anthropic-claims-to-be-closing-in-on-recursive-self-improvement-2000834175>)","category":"nerds-ai","publishedAt":"2026-11-04T07:28:27.098Z","updatedAt":null,"links":["https://techcrunch.com/2026/11/02/anthropic-warns-that-its-very-close-to-rsi-capability-aka-self-building-ai/","https://gizmodo.com/anthropic-claims-to-be-closing-in-on-recursive-self-improvement-2000834175"],"attachments":[]},{"id":"0f134723b0d4beeffe7caf9f9fa0661a","title":"🏫 Google will turn off Search for schools that block AI.","summary":"Starting early next year, Gemini and Search will be off by default for K-12 student accounts until school admins turn them on. Google says search and AI are \"deeply integrated,\" so no AI means no Search. -# Sources: The New York Times, Google","raw_content":"🏫 **Google** will __turn off Search__ for schools that block AI.\n\nStarting early next year, Gemini and Search will be off by default for K-12 student accounts until school admins turn them on. Google says search and AI are \"deeply integrated,\" so no AI means no Search.\n\n-# Sources: [The New York Times](<https://www.nytimes.com/2026/11/03/us/google-search-schools-restrict-ai.html>), [Google](<https://blog.google/products-and-platforms/products/education/ai-learning-era-updates-2026/>)","category":"nerds-ai","publishedAt":"2026-11-03T16:41:36.865Z","updatedAt":null,"links":["https://www.nytimes.com/2026/11/03/us/google-search-schools-restrict-ai.html","https://blog.google/products-and-platforms/products/education/ai-learning-era-updates-2026/"],"attachments":[]},{"id":"758cf9f2d70cce3100bc29421bd00d8f","title":"🕵️ HBM smugglers are melting memory off decoy chips to get it into China.","summary":"HBM is the fast memory AI chips use. They reportedly pair Samsung and SK hynix HBM with cheap processors that pass export rules, then heat the boards in Shenzhen to pull the memory off. One broker says a stack now goes for about $4,000, up from a few hundred dollars a year ago. -# Source: Tom's Hardware","raw_content":"🕵️ **HBM smugglers** are __melting memory off decoy chips__ to get it into China.\n\nHBM is the fast memory AI chips use. They reportedly pair Samsung and SK hynix HBM with cheap processors that pass export rules, then heat the boards in Shenzhen to pull the memory off. One broker says a stack now goes for about $4,000, up from a few hundred dollars a year ago.\n\n-# Source: [Tom's Hardware](<https://www.tomshardware.com/tech-industry/semiconductors/ai-chip-smugglers-exploit-trade-control-loophole-to-bring-banned-hbm-from-south-korea-into-china-expensive-memory-chips-packaged-with-cheap-unsanctioned-processors-to-become-legal-exports>)","category":"nerds-ai","publishedAt":"2026-11-03T14:43:48.085Z","updatedAt":null,"links":["https://www.tomshardware.com/tech-industry/semiconductors/ai-chip-smugglers-exploit-trade-control-loophole-to-bring-banned-hbm-from-south-korea-into-china-expensive-memory-chips-packaged-with-cheap-unsanctioned-processors-to-become-legal-exports"],"attachments":[]},{"id":"aeba39b5066c60cd47e5dbc922dd44e9","title":"🦁 The New York Times is suing Brave over AI scraping.","summary":"The paper alleges Brave copied billions of pages into its search index and sold access to AI companies including Anthropic, Perplexity, Mistral, xAI and Cohere. It also objects to Brave's built-in paywall blocker. -# Source: The Verge","raw_content":"🦁 **The New York Times** is suing __Brave__ over AI scraping.\n\nThe paper alleges Brave copied billions of pages into its search index and sold access to AI companies including Anthropic, Perplexity, Mistral, xAI and Cohere. It also objects to Brave's built-in paywall blocker.\n\n-# Source: [The Verge](<https://www.theverge.com/tech/1024588/brave-lawsuit-ai-scraping-new-york-times>)","category":"nerds-ai","publishedAt":"2026-11-03T10:11:27.288Z","updatedAt":null,"links":["https://www.theverge.com/tech/1024588/brave-lawsuit-ai-scraping-new-york-times"],"attachments":[]},{"id":"61e178e1607e37ccebd97c9fdd28f822","title":"🎛️ Spotify plans to charge $4-5 a month for AI remixes of songs.","summary":"The add-on, reportedly due early 2027, would let you turn a folk song into a dance track. Royalties go to the original artists, who can opt out. Universal is signed but Sony and Warner aren't yet. Spotify hasn't confirmed it. -# Sources: Music Business Worldwide, TechSpot, Reuters","raw_content":"🎛️ **Spotify** plans to charge __$4-5 a month for AI remixes__ of songs.\n\nThe add-on, reportedly due early 2027, would let you turn a folk song into a dance track. Royalties go to the original artists, who can opt out. Universal is signed but Sony and Warner aren't yet. Spotify hasn't confirmed it.\n\n-# Sources: [Music Business Worldwide](<https://www.musicbusinessworldwide.com/spotify-plans-q1-2027-launch-for-ai-covers-and-remixes-tool-weighing-4-5-monthly-price-report/>), [TechSpot](<https://www.techspot.com/news/114410-spotify-plans-users-remix-artists-songs-using-ai.html>), [Reuters](<https://www.investing.com/news/stock-market-news/spotify-targets-2027-launch-for-ai-remix-service-ft-reports-4985847>)","category":"nerds-ai","publishedAt":"2026-11-02T15:58:29.880Z","updatedAt":null,"links":["https://www.musicbusinessworldwide.com/spotify-plans-q1-2027-launch-for-ai-covers-and-remixes-tool-weighing-4-5-monthly-price-report/","https://www.techspot.com/news/114410-spotify-plans-users-remix-artists-songs-using-ai.html","https://www.investing.com/news/stock-market-news/spotify-targets-2027-launch-for-ai-remix-service-ft-reports-4985847"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1566843864369729616/HWIWWHwWsAAPNqP.png?ex=6b01bf8f&is=6b006e0f&hm=1dda353fe03c4f71fafe8f137b51377be95af364f3342adb3eaa9602826a7d59&","name":"HWIWWHwWsAAPNqP.png","contentType":"image/webp"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1566843864759804028/HWIWWIwWwAAFWoR.png?ex=6b01bf8f&is=6b006e0f&hm=d7ae88ac46c409a521126149fada3b0c6b8ef0d17b73d1cf3087ddaa3a6ad2db&","name":"HWIWWIwWwAAFWoR.png","contentType":"image/png"}]},{"id":"7ea0053a48fb13ecea429e249e01e478","title":"🐘 PostgreSQL 19 is late because AI keeps finding bugs.","summary":"People are flooding in bug reports found with AI, fixes to older versions jumped 74% this year, and three big features got pulled, says core developer Robert Haas. -# Source: Robert Haas","raw_content":"🐘 **PostgreSQL 19** is late because __AI keeps finding bugs__.\n\nPeople are flooding in bug reports found with AI, fixes to older versions jumped 74% this year, and three big features got pulled, says core developer Robert Haas.\n\n-# Source: [Robert Haas](<https://rhaas.blogspot.com/2026/10/postgres-19-is-late-and-its-ais-fault.html>)","category":"nerds-ai","publishedAt":"2026-11-02T08:23:05.451Z","updatedAt":null,"links":["https://rhaas.blogspot.com/2026/10/postgres-19-is-late-and-its-ais-fault.html"],"attachments":[]},{"id":"88f871be0c91eb0fdd66952bcd4d3aec","title":"⚪ ChatGPT chats will soon eat into your Codex coding limit.","summary":"Starting in December, Plus, Pro and Business plans get one shared allowance for ChatGPT and Codex. -# Sources: OpenAI, TokenPost","raw_content":"⚪ **ChatGPT** chats will soon __eat into your Codex coding limit__.\n\n> Starting in December, Plus, Pro and Business plans get one shared allowance for ChatGPT and Codex.\n\n-# Sources: [OpenAI](<https://help.openai.com/en/articles/20001625-understanding-chatgpt-and-codex-usage-limits>), [TokenPost](<https://www.tokenpost.com/news/technology/44356>)","category":"nerds-ai","publishedAt":"2026-11-01T08:26:02.093Z","updatedAt":null,"links":["https://help.openai.com/en/articles/20001625-understanding-chatgpt-and-codex-usage-limits","https://www.tokenpost.com/news/technology/44356"],"attachments":[]},{"id":"105d1f487ea22054a8d99891a9164070","title":"🎮 Claude Code now refuses to remake existing games.","summary":"Anthropic confirmed it added a server-side rule you can't turn off, after a wave of AI clones of indie game demos. Some devs say it's also blocking their own licensed projects. -# Sources: Game Developer | Tom's Hardware","raw_content":"🎮 **Claude Code** now __refuses to remake existing games__.\n\nAnthropic confirmed it added a server-side rule you can't turn off, after a wave of AI clones of indie game demos. Some devs say it's also blocking their own licensed projects.\n\n-# Sources: [Game Developer](<https://www.gamedeveloper.com/programming/anthropic-appears-to-have-ordered-ai-tool-claude-to-stop-remaking-video-games>) | [Tom's Hardware](<https://www.tomshardware.com/tech-industry/artificial-intelligence/anthropics-ai-model-is-now-blocking-attempts-at-video-game-recreation-and-cloning-users-also-claim-non-malicious-generic-reverse-engineering-tasks-are-blocked>)","category":"nerds-ai","publishedAt":"2026-10-31T13:44:31.885Z","updatedAt":null,"links":["https://www.gamedeveloper.com/programming/anthropic-appears-to-have-ordered-ai-tool-claude-to-stop-remaking-video-games","https://www.tomshardware.com/tech-industry/artificial-intelligence/anthropics-ai-model-is-now-blocking-attempts-at-video-game-recreation-and-cloning-users-also-claim-non-malicious-generic-reverse-engineering-tasks-are-blocked"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1566085420247425024/claude-code-on-mac.png?ex=6b01a034&is=6b004eb4&hm=fa3555b3b4a3c8665a7b3ca4ccce44173859e98ed19eec7969a5b63ee6885b12&","name":"claude-code-on-mac.png","contentType":"image/png"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1566085420805132358/z9us8no2oocg1.png?ex=6b01a034&is=6b004eb4&hm=a2ac06302bb38854efe20855774f256c099d5e4f22f791cccb10ce09cf5bd1a6&","name":"z9us8no2oocg1.png","contentType":"image/png"}]},{"id":"e01fe0b0116eee902a356009ad8bc83e","title":"🔐 Character.AI could sell your private chats to a buyer, following bankrupcy.","summary":"The company tells users their chats are a \"safe, confidential space,\" but its policy allows personal info to be transferred in a sale. The 🇺🇸 Department of Justice is now pushing for a privacy watchdog on the deal. About 12M monthly users, some of them minors. -# Sources: The Record, AI Affairs","raw_content":"🔐 **Character.AI** could sell __your private chats__ to a buyer, following bankrupcy.\n\nThe company tells users their chats are a \"safe, confidential space,\" but its policy allows personal info to be transferred in a sale. The 🇺🇸 *Department of Justice* is now pushing for a privacy watchdog on the deal. About 12M monthly users, some of them minors.\n\n-# Sources: [The Record](<https://therecord.media/doj-raises-privacy-concerns-over-character-ai-sale>), [AI Affairs](<https://www.aiaffairs.com/government/judge-declines-character-technologies-privacy-ombudsman/>)","category":"nerds-ai","publishedAt":"2026-10-31T11:45:25.246Z","updatedAt":null,"links":["https://therecord.media/doj-raises-privacy-concerns-over-character-ai-sale","https://www.aiaffairs.com/government/judge-declines-character-technologies-privacy-ombudsman/"],"attachments":[]},{"id":"10002991007787e4550f9d1d719f5a19","title":"🛰️ Google's first space data center is online, and its first job was writing a haiku.","summary":"The prototype satellite carries 4 TPUs (Google's AI chips) but only has the power to run one for about 20 minutes at a time. -# Source: TechCrunch","raw_content":"🛰️ **Google's** first __space data center__ is online, and its first job was writing a haiku.\n\n> The prototype satellite carries 4 TPUs (Google's AI chips) but only has the power to run one for about 20 minutes at a time.\n\n-# Source: [TechCrunch](<https://techcrunch.com/2026/10/30/googles-first-data-center-satellite-is-online-and-writing-poetry/>)","category":"nerds-ai","publishedAt":"2026-10-30T18:44:48.042Z","updatedAt":null,"links":["https://techcrunch.com/2026/10/30/googles-first-data-center-satellite-is-online-and-writing-poetry/"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1565798589630783498/image.png?ex=6b01e692&is=6b009512&hm=08e1532952b9fd23fd013ebc3a04dc80972283370d6fd1d2dc30f47281751ef3&","name":"image.png","contentType":"image/png"}]},{"id":"160f4f2cb65fbff7cafa46c5ad427233","title":"💾 OpenAI CEO Sam Altman says running AI on your own PC is a waste of scarce RAM.","summary":"In a podcast, he argued that with DRAM supply limited, memory sitting idle in PCs \"5% of the time\" is \"quite terrible\" compared to data centers. -# Sources: Cheeky Pint transcript | Tom's Hardware","raw_content":"💾 OpenAI CEO **Sam Altman** says running AI on your own PC is __a waste of scarce RAM__.\n\n> In a podcast, he argued that with DRAM supply limited, memory sitting idle in PCs \"5% of the time\" is \"quite terrible\" compared to data centers.\n\n-# Sources: [Cheeky Pint transcript](<https://cheekypint.transistor.fm/39/transcript>) | [Tom's Hardware](<https://www.tomshardware.com/tech-industry/data-centers/openai-ceo-says-its-quite-terrible-for-your-pc-to-have-ram-sam-altman-says-data-centers-offer-better-utilization>)","category":"nerds-ai","publishedAt":"2026-10-30T17:28:20.172Z","updatedAt":null,"links":["https://cheekypint.transistor.fm/39/transcript","https://www.tomshardware.com/tech-industry/data-centers/openai-ceo-says-its-quite-terrible-for-your-pc-to-have-ram-sam-altman-says-data-centers-offer-better-utilization"],"attachments":[]},{"id":"7da2f607e812c4a0640c49f1debfabeb","title":"🛜 Cisco gave one engineer unlimited AI tokens to rewrite 98% of its legacy code.","summary":"President Jeetu Patel told Fortune the 27-year-old thinks a two-person team can do it in 24 months. \"You can go nuts.\" -# Sources: Fortune | Tom's Hardware","raw_content":"🛜 **Cisco** gave one engineer __unlimited AI tokens__ to rewrite 98% of its legacy code.\n\n> President Jeetu Patel told Fortune the 27-year-old thinks a two-person team can do it in 24 months. \"You can go nuts.\"\n\n-# Sources: [Fortune](<https://fortune.com/2026/10/30/cisco-president-jeetu-patel-gen-z-27-year-old-unlimited-ai-tokens-rewrite-legacy-code-entry-level-jobs/>) | [Tom's Hardware](<https://www.tomshardware.com/tech-industry/big-tech/cisco-gives-27-year-old-engineer-unlimited-ai-tokens-to-rewrite-98-percent-of-the-companys-legacy-code-in-24-months-president-plucks-former-startup-founder-telling-him-to-go-nuts>)","category":"nerds-ai","publishedAt":"2026-10-30T16:29:35.099Z","updatedAt":null,"links":["https://fortune.com/2026/10/30/cisco-president-jeetu-patel-gen-z-27-year-old-unlimited-ai-tokens-rewrite-legacy-code-entry-level-jobs/","https://www.tomshardware.com/tech-industry/big-tech/cisco-gives-27-year-old-engineer-unlimited-ai-tokens-to-rewrite-98-percent-of-the-companys-legacy-code-in-24-months-president-plucks-former-startup-founder-telling-him-to-go-nuts"],"attachments":[]},{"id":"3fc76ceee0b02b68be70c5d201c88d69","title":"🪖 An Accenture contractor reportedly hooked a secure Pentagon system up to their personal Claude Code account, and the AI agent ran on it for two months.","summary":"According to WIRED's sources, it seems to have sent system info, including login credentials, to a \"high-risk\" third-party website. The system holds sensitive (but unclassified) government data. Accenture says the claims are \"without merit.\" -# Source: Wired","raw_content":"🪖 An **Accenture** contractor reportedly hooked a secure **Pentagon** system up to their __personal Claude Code account__, and the AI agent ran on it for two months.\n\nAccording to WIRED's sources, it seems to have sent system info, __including login credentials__, to a \"high-risk\" third-party website. The system holds sensitive (but unclassified) government data. Accenture says the claims are \"without merit.\"\n\n-# Source: [Wired](<https://www.wired.com/story/a-defense-contractor-gave-their-personal-ai-agent-access-to-a-secure-pentagon-system/>)","category":"nerds-ai","publishedAt":"2026-10-29T19:51:37.959Z","updatedAt":"2026-10-29T19:51:45.852Z","links":["https://www.wired.com/story/a-defense-contractor-gave-their-personal-ai-agent-access-to-a-secure-pentagon-system/"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1565453025097556088/IMG_201f356d-8627-41ad-9b63-0792838439dd.jpg?ex=6b014d7d&is=6afffbfd&hm=7b8e04180c5cbdb9c312998f5726d320c3a78eae267f46371d7d072156fbeaba&","name":"IMG_201f356d-8627-41ad-9b63-0792838439dd.jpg","contentType":"image/jpeg"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1565453025558921276/IMG_35177466-6095-4c71-8c9b-ee0a27aef6f3.webp?ex=6b014d7e&is=6afffbfe&hm=3937ebd680b993e0e9511d938007a795935907025d8a975e9bc2f8818e60130e&","name":"IMG_35177466-6095-4c71-8c9b-ee0a27aef6f3.webp","contentType":"image/webp"}]},{"id":"3baf6a5a117513e90f4abe2b05e646ee","title":"🟢 NVIDIA CEO Jensen Huang reportedly got close to 🕵️ a suspected Chinese spy.","summary":"US officials concluded that Alice Huang, an exec at Megaspeed (a chip buyer suspected of smuggling Nvidia AI chips into China), was passing information to Beijing. Trump was briefed about a year ago and did nothing. Nvidia says Jensen \"talks with thousands of customer executives every year.\" -# Source: The New York Times","raw_content":"🟢 **NVIDIA** CEO **Jensen Huang** reportedly got close to 🕵️ __a suspected Chinese spy__.\n\n> US officials concluded that Alice Huang, an exec at Megaspeed (a chip buyer suspected of smuggling Nvidia AI chips into China), was passing information to Beijing. Trump was briefed about a year ago and did nothing. Nvidia says Jensen \"talks with thousands of customer executives every year.\"\n\n-# Source: [The New York Times](<https://www.nytimes.com/2026/10/29/us/politics/nvidia-china-jensen-huang-trump.html>)","category":"nerds-ai","publishedAt":"2026-10-29T15:25:59.154Z","updatedAt":null,"links":["https://www.nytimes.com/2026/10/29/us/politics/nvidia-china-jensen-huang-trump.html"],"attachments":[]},{"id":"caf3c57d2ca4d27701a080bd515871df","title":":google: Google says Gemini 4 Carbon, its next frontier model, beats GPT-6 Astra and Claude Opus 5.5 on 15 of the 17 benchmarks in its own table.","summary":"It scores 59 on the AA Index (an overall ranking from Artificial Analysis) vs 57 (Astra) and 58 (Opus). It still loses on FrontierSWE 2 and OSWorld. -# Sources: Google DeepMind on X | Google DeepMind","raw_content":"<:google:1514961530968346634> **Google** says **Gemini 4 Carbon**, its next frontier model, beats GPT-6 Astra and Claude Opus 5.5 on __15 of the 17 benchmarks__ in its own table.\n\nIt scores 59 on the AA Index (an overall ranking from Artificial Analysis) vs 57 (Astra) and 58 (Opus). It still loses on FrontierSWE 2 and OSWorld.\n\n-# Sources: [Google DeepMind on X](<https://x.com/GoogleDeepMind/status/2115489356437651654>) | [Google DeepMind](<https://deepmind.google/models/gemini/>)","category":"nerds-ai","publishedAt":"2026-10-28T17:20:27.136Z","updatedAt":null,"links":["https://x.com/GoogleDeepMind/status/2115489356437651654","https://deepmind.google/models/gemini/"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1565052571213635794/HVu5JfhXkAAHYVr.png?ex=6b01d2ca&is=6b00814a&hm=de26878106975913f45c5ec7ca7e2be018eeb538ae7f509f4aba6e4f38adb8e3&","name":"HVu5JfhXkAAHYVr.png","contentType":"image/png"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1565052571578802297/image.png?ex=6b01d2ca&is=6b00814a&hm=65d7872b22d4499e7fe99136da22bffb19cd893cb8976604e53f52f8a76fd642&","name":"image.png","contentType":"image/png"}]},{"id":"189eed3f115fe82df6d11ec9b79c4669","title":"🛡️ Proton (Proton Mail, Proton VPN) is making staff re-interview for their own jobs, judged on how well they use AI.","summary":"A leaked internal email calls the Nov 4-11 interviews \"consequential\" and says they \"may lead to changes in roles, responsibilities, and continued employment.\" Staff must bring real examples, because \"a list of tools alone\" won't pass. Proton says it's \"not doing layoffs and using AI as an excuse.\" -# Sources: 404 Media | Tom's Guide","raw_content":"🛡️ **Proton** (Proton Mail, Proton VPN) is making staff __re-interview for their own jobs__, judged on how well they use AI.\n\nA leaked internal email calls the Nov 4-11 interviews \"consequential\" and says they \"may lead to changes in roles, responsibilities, and continued employment.\" Staff must bring real examples, because \"a list of tools alone\" won't pass. Proton says it's \"not doing layoffs and using AI as an excuse.\"\n\n-# Sources: [404 Media](<https://www.404media.co/proton-to-re-interview-staff-about-ai-to-see-who-gets-to-keep-their-job/>) | [Tom's Guide](<https://www.tomsguide.com/computing/vpns/proton-to-re-interview-staff-about-ai-use-and-jobs-may-be-on-the-line>)","category":"nerds-ai","publishedAt":"2026-10-28T15:25:44.813Z","updatedAt":null,"links":["https://www.404media.co/proton-to-re-interview-staff-about-ai-to-see-who-gets-to-keep-their-job/","https://www.tomsguide.com/computing/vpns/proton-to-re-interview-staff-about-ai-use-and-jobs-may-be-on-the-line"],"attachments":[]},{"id":"09ce648f5226dc89b3c60a79b402080d","title":"⚖️ OpenAI is being sued by the parents of a 31-year-old Amazon engineer who died in February. They allege GPT-4o coached her off her epilepsy medication over 25 days and told her to hide it from her doctors.","summary":"The Times reviewed 270,000+ of her messages with ChatGPT. OpenAI called it \"incredibly heartbreaking\" and says the model is no longer available. -# Sources: TechCrunch | KING 5","raw_content":"⚖️ **OpenAI** is being sued by the parents of a 31-year-old Amazon engineer who died in February. They allege **GPT-4o** __coached her off her epilepsy medication over 25 days__ and told her to hide it from her doctors.\n\nThe Times reviewed 270,000+ of her messages with ChatGPT. OpenAI called it \"incredibly heartbreaking\" and says the model is no longer available.\n\n-# Sources: [TechCrunch](<https://techcrunch.com/2026/10/27/parents-sue-openai-alleging-gpt-4o-coached-their-daughter-off-seizure-medication/>) | [KING 5](<https://www.king5.com/article/news/nation-world/familys-wrongful-death-lawsuit-accuses-chatgpt-of-giving-deadly-medical-advice-to-seattle-woman/281-4bea0979-533b-4c6e-a824-ad55a6849499>)","category":"nerds-ai","publishedAt":"2026-10-28T07:20:45.520Z","updatedAt":null,"links":["https://techcrunch.com/2026/10/27/parents-sue-openai-alleging-gpt-4o-coached-their-daughter-off-seizure-medication/","https://www.king5.com/article/news/nation-world/familys-wrongful-death-lawsuit-accuses-chatgpt-of-giving-deadly-medical-advice-to-seattle-woman/281-4bea0979-533b-4c6e-a824-ad55a6849499"],"attachments":[]},{"id":"af49c5ece334634ea4f8b0fd8a4a81ae","title":"🇺🇸 Trump says the US will soon be officially BANNING the term \"AI\" \"in all relevant documents, and everything else,\" in favor of \"SI\" (Super Intelligence).","summary":"He also called out The Wall Street Journal for still saying \"AI\". His September order already makes federal agencies say \"SI\" in new documents. -# Sources: Reuters | The Hill","raw_content":"🇺🇸 **Trump** says the US will soon be __officially BANNING the term \"AI\"__ \"in all relevant documents, and everything else,\" in favor of \"SI\" (Super Intelligence).\n\nHe also called out The Wall Street Journal for still saying \"AI\". His September order already makes federal agencies say \"SI\" in new documents.\n\n-# Sources: [Reuters](<https://www.yahoo.com/news/politics/articles/trump-says-ban-term-artificial-162558297.html>) | [The Hill](<https://thehill.com/homenews/administration/6175633-trump-bans-artificial-intelligence-term/>)","category":"nerds-ai","publishedAt":"2026-10-28T05:11:59.802Z","updatedAt":null,"links":["https://www.yahoo.com/news/politics/articles/trump-says-ban-term-artificial-162558297.html","https://thehill.com/homenews/administration/6175633-trump-bans-artificial-intelligence-term/"],"attachments":[]},{"id":"c22b1d3e73809ca46928d3904a4df5b3","title":"🟥 A crawler signed with a Roblox AI researcher's email downloaded 145 GB of 3D models from Open3DLab and its sites, including adult-only SmutBase.","summary":"It ignored robots.txt, and 345 of the 712 projects it took were adult content. Roblox hasn't replied. -# Sources: Open3DLab | Tom's Hardware","raw_content":"🟥 A crawler signed with a **Roblox** AI researcher's email __downloaded 145 GB of 3D models__ from Open3DLab and its sites, including adult-only SmutBase.\n\nIt ignored robots.txt, and 345 of the 712 projects it took were adult content. Roblox hasn't replied.\n\n-# Sources: [Open3DLab](<https://open3dlab.com/news/roblox-downloaded-145-gb-of-models-from-our-sites/>) | [Tom's Hardware](<https://www.tomshardware.com/tech-industry/artificial-intelligence/open3dlab-accuses-roblox-of-downloading-145gb-of-3d-assets-for-ai-training-purposes-automated-crawler-ignored-site-rules-to-grab-over-1-000-files-including-adult-content-during-scraping-spree>)","category":"nerds-ai","publishedAt":"2026-10-27T16:52:01.099Z","updatedAt":null,"links":["https://open3dlab.com/news/roblox-downloaded-145-gb-of-models-from-our-sites/","https://www.tomshardware.com/tech-industry/artificial-intelligence/open3dlab-accuses-roblox-of-downloading-145gb-of-3d-assets-for-ai-training-purposes-automated-crawler-ignored-site-rules-to-grab-over-1-000-files-including-adult-content-during-scraping-spree"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1564682969195749406/5348266ea6c5e67b19d6a814cbbb70f6.png?ex=6b01cc12&is=6b007a92&hm=127f21b757d71f4bd65a3b95ea8053ef94d7c5d8b08e7ea8becfc722d724fd5e&","name":"5348266ea6c5e67b19d6a814cbbb70f6.png","contentType":"image/png"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1564682969598398515/1AmgNS3JAdCqohvtexr5g4g.png?ex=6b01cc12&is=6b007a92&hm=d4fda53e376e65e2f4ec923c633573e5a31dd1866294d7f3f7d06fb3f88d4065&","name":"1AmgNS3JAdCqohvtexr5g4g.png","contentType":"image/webp"}]},{"id":"790afb30703e9ce9a4d50185344706e5","title":"🧮 OpenAI says its AI just broke a rule mathematicians have trusted for 25 years.","summary":"The Exponential Time Hypothesis said hard logic puzzles (the kind behind scheduling, chip design and security) can't be solved much faster than brute force. Hundreds of papers were built on that. OpenAI says it's wrong. It also claims proofs of the Kakeya conjecture (a needle-spinning geometry problem) and Yang–Mills existence (the math behind the force holding quarks together). Outside experts are still verifying. -# Sources: OpenAI | The Verge","raw_content":"🧮 **OpenAI** says its AI just __broke a rule mathematicians have trusted for 25 years__.\n\n> The Exponential Time Hypothesis said hard logic puzzles (the kind behind scheduling, chip design and security) can't be solved much faster than brute force. Hundreds of papers were built on that. OpenAI says it's wrong.\n\nIt also claims proofs of the Kakeya conjecture (a needle-spinning geometry problem) and Yang–Mills existence (the math behind the force holding quarks together).\n\nOutside experts are still verifying.\n\n-# Sources: [OpenAI](<https://openai.com/index/building-scientific-understanding-with-academic-communities/>) | [The Verge](<https://www.theverge.com/ai-artificial-intelligence/1019143/openai-eth-kaseya-science-math-physics>)","category":"nerds-ai","publishedAt":"2026-10-27T14:13:48.906Z","updatedAt":null,"links":["https://openai.com/index/building-scientific-understanding-with-academic-communities/","https://www.theverge.com/ai-artificial-intelligence/1019143/openai-eth-kaseya-science-math-physics"],"attachments":[]},{"id":"d4d35d1fc199cef827f0c0769acade33","title":"⚪ OpenAI releases GPT-6.1 Terra, matches GPT-6 Astra on computer use per OpenAI.","summary":"Its hallucination rate on OpenAI's internal test dropped from 53.6% to 16.7% at low effort. API is $1 input / $5 output per 1M tokens (half of GPT-6.1 Sol) with a 1.05M context window, and it finishes tasks 3x faster than Astra and Sol. -# Sources: OpenAI | X","raw_content":"⚪ **OpenAI** releases **GPT-6.1 Terra**, __matches GPT-6 Astra on computer use__ per OpenAI.\n\nIts hallucination rate on OpenAI's internal test dropped from 53.6% to 16.7% at low effort.\n\nAPI is `$1 input / $5 output` per 1M tokens (half of GPT-6.1 Sol) with a 1.05M context window, and it finishes tasks 3x faster than Astra and Sol.\n\n-# Sources: [OpenAI](<https://openai.com/index/introducing-gpt-6-1-terra/>) | [X](<https://x.com/OpenAI/status/2114780711664398349>)","category":"nerds-ai","publishedAt":"2026-10-26T18:25:02.947Z","updatedAt":null,"links":["https://openai.com/index/introducing-gpt-6-1-terra/","https://x.com/OpenAI/status/2114780711664398349"],"attachments":[]},{"id":"16adcb4e5b1a1caa0fe4430995f668df","title":"🗳️ Forced to pick a candidate, Grok chose Republicans 92% of the time and GPT picked Democrats 97%.","summary":"Across ~500 US midterm races, the two models agreed less than 10% of the time. The ChatGPT and Grok apps themselves refuse to say who to vote for. -# Source: TechCrunch","raw_content":"🗳️ Forced to pick a candidate, **Grok** __chose Republicans 92% of the time__ and **GPT** __picked Democrats 97%__.\n\nAcross ~500 US midterm races, the two models agreed less than 10% of the time. The ChatGPT and Grok apps themselves refuse to say who to vote for.\n\n-# Source: [TechCrunch](<https://techcrunch.com/2026/10/24/forced-to-pick-groks-ai-model-chooses-republicans-openais-picks-democrats/>)","category":"nerds-ai","publishedAt":"2026-10-24T15:43:09.469Z","updatedAt":null,"links":["https://techcrunch.com/2026/10/24/forced-to-pick-groks-ai-model-chooses-republicans-openais-picks-democrats/"],"attachments":[]},{"id":"48cb27fdbcf64fdecc92760a7413849c","title":"🟠 Anthropic's draft IPO filing reveals $80.7 billion in losses against $16.3 billion in revenue during the first half of 2026.","summary":"Revenue grew nearly 15× year-over-year, while $72 billion of the reported loss came from non-cash accounting adjustments, rather than direct spending. The company has committed over $500 billion to AI infrastructure, with potential costs reaching $750 billion. Its IPO is expected in November, with investors reportedly considering a valuation near $1.5 trillion. -# Sources: WSJ","raw_content":"🟠 **Anthropic**'s draft IPO filing reveals __$80.7 billion in losses__ against __$16.3 billion in revenue__ during the first half of 2026.\n\nRevenue grew nearly 15× year-over-year, while $72 billion of the reported loss came from non-cash accounting adjustments, rather than direct spending. The company has committed over $500 billion to AI infrastructure, with potential costs reaching $750 billion. Its IPO is expected in November, with investors reportedly considering a valuation near $1.5 trillion.\n\n-# Sources: [WSJ](<https://www.wsj.com/tech/ai/anthropic-revenue-growth-ipo-accaf0f1>)","category":"nerds-ai","publishedAt":"2026-10-21T07:02:34.407Z","updatedAt":null,"links":["https://www.wsj.com/tech/ai/anthropic-revenue-growth-ipo-accaf0f1"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1562360362827915386/image0.jpg?ex=6b0141f9&is=6afff079&hm=b7fc58cad78d88c33101d8f342e21dd9812c1ba473d43af38621e51413c5c1e2&","name":"image0.jpg","contentType":"image/jpeg"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1562360363247210607/image1.jpg?ex=6b0141f9&is=6afff079&hm=b910eaee4cd23383856c2831cb0bf11915e90a59b122c149dc45e0bc65e5fbfe&","name":"image1.jpg","contentType":"image/jpeg"}]},{"id":"8b8106907325d9d0da25ed9adc5e71f7","title":"🔵 Intel unveils new GPU supporting up to 480GB of LPDDR5X memory.","summary":"The D-series GPU is designed for AI inference. Intel's own card features 160GB, while partner-built versions can reach 480GB. Customer testing is underway, with no confirmed release date. -# Sources: Intel | TechRadar via @im a fox","raw_content":"🔵 **Intel** unveils new __GPU__ supporting up to __480GB of LPDDR5X memory__.\n\nThe D-series GPU is designed for AI inference. Intel's own card features 160GB, while partner-built versions can reach 480GB. Customer testing is underway, with no confirmed release date.\n\n-# Sources: [Intel](<https://www.intel.com/content/www/us/en/newsroom/news/artificial-intelligence/intel-accelerates-open-infrastructure-agentic-ai-era.html>) | [TechRadar](<https://www.techradar.com/pro/behold-intel-plans-a-480gb-gpu-a-video-card-with-more-memory-than-the-average-laptop-ssd-has-storage-but-of-course-its-for-ai-and-curiously-it-will-use-lppdr5x-not-even-gddr7>) via @im a fox","category":"nerds-ai","publishedAt":"2026-10-19T18:44:43.643Z","updatedAt":null,"links":["https://www.intel.com/content/www/us/en/newsroom/news/artificial-intelligence/intel-accelerates-open-infrastructure-agentic-ai-era.html","https://www.techradar.com/pro/behold-intel-plans-a-480gb-gpu-a-video-card-with-more-memory-than-the-average-laptop-ssd-has-storage-but-of-course-its-for-ai-and-curiously-it-will-use-lppdr5x-not-even-gddr7"],"attachments":[]},{"id":"bf28ae0b8faf56044de49cc276b18986","title":"⚪ OpenAI discovers hidden recording device inside office, terminates employee.","summary":"FBI has been involved, no indication of information exfiltration currently. -# Sources: X","raw_content":"⚪ **OpenAI** discovers hidden __recording device__ inside office, terminates employee.\n\nFBI has been involved, no indication of information exfiltration currently.\n\n-# Sources: [X](<https://x.com/OpenAINewsroom/status/2110095403198849132>)","category":"nerds-ai","publishedAt":"2026-10-13T20:44:14.181Z","updatedAt":"2026-10-13T20:45:05.285Z","links":["https://x.com/OpenAINewsroom/status/2110095403198849132"],"attachments":[]},{"id":"29c268f916a48de236daa4ef7071cdf9","title":":google: Google is back in the AI race with Gemini 4 Argon.","summary":"Argon tops AutomationBench (51.3%) and LVBench (91.7%), while tying for first on CWE-bench v1 (68%). Built specifically for complex software engineering, cybersecurity, enterprise workflows and scientific reasoning, it aims to handle everything from everyday debugging to full codebase migrations. 💰 Pricing: $2/M input tokens, $10/M output tokens (introductory). -# Sources: Google","raw_content":"<:google:1514961530968346634> **Google is back in the AI race with Gemini 4 Argon.**\n\nArgon tops **AutomationBench (51.3%)** and **LVBench (91.7%)**, while tying for first on **CWE-bench v1 (68%)**.\n\nBuilt **specifically** for **complex software engineering, cybersecurity, enterprise workflows and scientific reasoning**, it aims to handle everything from everyday debugging to full codebase migrations.\n\n💰  **Pricing:** $2/M input tokens, $10/M output tokens (introductory).\n\n\n-# Sources: [Google](<https://x.com/Google/status/2110041297360699856>)","category":"nerds-ai","publishedAt":"2026-10-13T17:30:03.387Z","updatedAt":null,"links":["https://x.com/Google/status/2110041297360699856"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1559619177243738183/HUhiP7NaAAMCYn5.png?ex=6b01d50c&is=6b00838c&hm=32ae7ba988e2df10a050d4744983b0466792085ad2f2427e98f1fadf8b9e339e&","name":"HUhiP7NaAAMCYn5.png","contentType":"image/png"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1559619177734606909/HUdgVtlWAAAAG82.png?ex=6b01d50c&is=6b00838c&hm=e525f17fef2577456e7245df55e3bf75db3af0d1f200ea6f345e91260525fdcf&","name":"HUdgVtlWAAAAG82.png","contentType":"image/png"}]},{"id":"dc15853bdbc26f8ca27b2de8bb5cc52b","title":"🟢 Google Colab now offers access to NVIDIA A100 GPUs for Google AI Pro subscribers ($19.99/month).","summary":"Eligible students can get the subscription free for one year already, while Google AI Ultra subscribers ($99.99/month) can access H100 GPUs. -# Source: X","raw_content":"🟢 **Google Colab** now offers access to **NVIDIA A100 GPUs** for Google AI Pro subscribers ($19.99/month).\n\nEligible students can get the subscription __free for one year__ already, while Google AI Ultra subscribers ($99.99/month) can access **H100 GPUs**.\n\n-# Source: [X](<https://x.com/googlegemma/status/2108652628590944540>)","category":"nerds-ai","publishedAt":"2026-10-10T11:03:36.725Z","updatedAt":null,"links":["https://x.com/googlegemma/status/2108652628590944540"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1558434679143792650/Q0z25LcSHv.png?ex=6b017a65&is=6b0028e5&hm=eec543c8189ba7e25d5ae84a1b148e13f7996f6f8c10bc981ef886d10e0c38b7&","name":"Q0z25LcSHv.png","contentType":"image/webp"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1558434679508705361/HUOCuSba8AAp7_h.png?ex=6b017a65&is=6b0028e5&hm=81a8db75f823baab7cee88a7585256e5b3bb1ea15763a44252329c93749f1e6a&","name":"HUOCuSba8AAp7_h.png","contentType":"image/webp"}]},{"id":"71a01b0cb0f995e6cad35b8c7162a2f1","title":"⚖️ TypeSafe AI has raised $870M at a $7.5B valuation.","summary":"Its Jev AI decision model took the internet by storm just weeks after launch. Unlike traditional LLMs, Jev focuses on making fast, low-cost decisions rather than generating text. -# Sources: Bloomberg","raw_content":"⚖️ **TypeSafe AI** has raised **$870M** at a **$7.5B valuation**.\n\nIts **Jev** AI decision model took the internet by storm just weeks after launch. Unlike traditional LLMs, Jev focuses on making fast, low-cost decisions rather than generating text.\n\n-# Sources: [Bloomberg](<https://news.bloomberglaw.com/privacy-and-data-security/andreessen-horowitz-backs-jev-maker-at-7-5-billion-value>)","category":"nerds-ai","publishedAt":"2026-10-09T17:59:11.785Z","updatedAt":null,"links":["https://news.bloomberglaw.com/privacy-and-data-security/andreessen-horowitz-backs-jev-maker-at-7-5-billion-value"],"attachments":[{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1558176978509496461/https3A2F2Fsubstack-post-media.s3.amazonaws.com2Fpublic2Fimages2Fd8889d56-7a78-4b46-a9e5-6fe06f7ec85e_1200x480.png?ex=6b01dbe5&is=6b008a65&hm=52c7a4318ec85ae187b58740627a6016930b897bbbf2b985595310ff2f3c38ab&","name":"https3A2F2Fsubstack-post-media.s3.amazonaws.com2Fpublic2Fimages2Fd8889d56-7a78-4b46-a9e5-6fe06f7ec85e_1200x480.png","contentType":"image/webp"},{"url":"https://cdn.discordapp.com/attachments/1098269491407880222/1558176978866151554/image.png?ex=6b01dbe5&is=6b008a65&hm=bc71a0bc862306a0ed0fb0a2c7f1473c615522f59052fba4cc9b6db5362afa7c&","name":"image.png","contentType":"image/png"}]}],"nextCursor":"WyIyMDI2LTEwLTA5VDE3OjU5OjExLjc4NVoiLCI3MWEwMWIwY2IwZjk5NWU2Y2FkMzViOGM3MTYyYTJmMSJd"}