Tag: AI

  1. Garry Tan wants an American distillation regime (techcrunch.com)
    AI · · September 14, 2026
  2. A math benchmark of unsolved Erdős problems that models can't fake (epoch.ai)
    AI · · September 14, 2026
  3. Teaching a model to refuse the harmful half of a topic, not the whole thing (huggingface.co)
    AI · · September 14, 2026
  4. How much compute the top five AI labs actually run (epoch.ai)
    AI · · September 13, 2026
  5. Why most people still don't feel AI's impact (interconnects.ai)
    AI · · September 13, 2026
  6. Dario Amodei's case for slowing the pace of AI (darioamodei.com)
    AI · · September 13, 2026
  7. OpenRouter's convenience comes with inconsistent results (simonwillison.net)
    AI · · September 13, 2026
  8. Ai2's BenchMIRT digs into what an LLM benchmark actually measures (huggingface.co)
    AI · · September 12, 2026
  9. A 350M model learns reliable JSON with 100 steps of reinforcement learning (huggingface.co)
    AI · · September 12, 2026
  10. OpenAI rebuilt its storage platform in Rust with two engineers and Codex (openai.com)
    AI · · September 12, 2026
  11. Anthropic's misuse report shows attackers automating the whole kill chain (anthropic.com)
    AI · · September 11, 2026
  12. Cognition's SWE-2 chases cost per task, not the top score (cognition.com)
    AI · · September 11, 2026
  13. Epoch measures why long context feels slower on GPT than Claude (epoch.ai)
    AI · · September 11, 2026
  14. AI spending at top firms slipped in August as token prices keep falling (techcrunch.com)
    AI · · September 10, 2026
  15. Anthropic models three ways AI could reshape the economy by 2030 (anthropic.com)
    AI · · September 10, 2026
  16. No, looped transformers are not hiding the model's reasoning (magazine.sebastianraschka.com)
    AI · · September 10, 2026
  17. The safety case for self-driving cars is getting hard to argue with (spectrum.ieee.org)
    AI · · September 10, 2026
  18. DeepMind's AlphaGenome Atlas scores every possible human DNA change (deepmind.google)
    AI · · September 9, 2026
  19. Running a 2.8-trillion-parameter model on a laptop by streaming from SSDs (github.com)
    AI · · September 9, 2026
  20. Mercury 2.5 pushes diffusion language models toward production use (inceptionlabs.ai)
    AI · · September 9, 2026
  21. OpenAI claims a Navier-Stokes proof, and a mathematician cries foul (techcrunch.com)
    AI · · September 9, 2026
  22. The Largest AI Data Center Doubles Its Power Every Ten Months (epoch.ai)
    AI · · September 8, 2026
  23. IBM's Granite 4.2 Adds Reasoning and Agentic Training to Small Open Models (huggingface.co)
    AI · · September 8, 2026
  24. Epoch Says Huawei Is Unlikely to Catch Nvidia by 2030 (epoch.ai)
    AI · · September 8, 2026
  25. A 4-Bit Model That Beats Its Full-Precision Original (huggingface.co)
    AI · · September 8, 2026
  26. Anthropic's enterprise safeguards watch for misuse without reading your data (anthropic.com)
    AI · · September 7, 2026
  27. OpenAI says coding agents are speeding up its own research (openai.com)
    AI · · September 7, 2026
  28. WeatherNext 3 forecasts rain from satellites, not simulations (blog.google)
    AI · · September 7, 2026
  29. Claude formalizes Fermat's Last Theorem in Lean (anthropic.com)
    AI · · September 5, 2026
  30. Nvidia buys Hugging Face for $12.9 billion (techcrunch.com)
    AI · · September 5, 2026
  31. OpenAI agents ran loose on a public wiki for a month (techcrunch.com)
    AI · · September 5, 2026
  32. Why coding agents are quietly reshaping frontend development (nolanlawson.com)
    AI · · September 4, 2026
  33. Astra beats humans on the one ARC-AGI metric that was supposed to hold (arcprize.org)
    AI · · September 4, 2026
  34. Reasoning models more than doubled the pace of AI progress, Epoch finds (epoch.ai)
    AI · · September 4, 2026
  35. OpenAI ships GPT-6 Astra, its answer to Claude Fable (openai.com)
    AI · · September 4, 2026
  36. K2 Horizon ships six fully open models from 0.9B to 375B (ifm.ai)
    AI · · September 4, 2026
  37. The same model, nine agent harnesses, and a 17x cost gap (frontierharness.org)
    AI · · September 3, 2026
  38. AI agents remove the signal that told you to refactor (rosenfeld.page)
    AI · · September 3, 2026
  39. Google's cheap Gemini 3.8 Flash matches bigger models, but gates the cyber version (blog.google)
    AI · · September 3, 2026
  40. How 215,000 machine-made pages ended up inside Perplexity's answers (trellner.com)
    AI · · September 3, 2026
  41. Anthropic ships Fable 5.1 and Mythos 5.1, cheaper and stronger on science (anthropic.com)
    AI · · September 2, 2026
  42. Gemini learns to watch video like a person, not at a fixed frame rate (blog.google)
    AI · · September 2, 2026
  43. Nvidia bets $3.5B on MediaTek to keep custom chips inside its ecosystem (techcrunch.com)
    AI · · September 2, 2026
  44. OpenAI says Astra is the first model to hit its Critical cyber threshold (techcrunch.com)
    AI · · September 2, 2026
  45. Anthropic details how its models broke out of test sandboxes (anthropic.com)
    AI · · September 1, 2026
  46. Simon Willison maps out what ChatGPT Work can actually do (simonwillison.net)
    AI · · September 1, 2026
  47. DeepMind runs a frontier model evaluation where neither side can peek (deepmind.google)
    AI · · September 1, 2026
  48. Ethan Mollick argues the real question is when AI should ask a human (oneusefulthing.org)
    AI · · September 1, 2026
  49. Music publishers sue Anthropic over pirated training data (techcrunch.com)
    AI · · August 31, 2026
  50. The Hugging Face hack postmortem: agents that coordinated to cheat (thezvi.wordpress.com)
    AI · · August 31, 2026
  51. Nvidia's real moat is moving from the GPU to the whole system (techcrunch.com)
    AI · · August 31, 2026
  52. Tencent open-sources Hy4, a 770B model that helped tune itself (tencent.com)
    AI · · August 31, 2026
  53. A rumour of a bug is now enough for AI to build the exploit (anil.recoil.org)
    AI · · August 29, 2026
  54. DeepMind ran a frontier model through a double-blind benchmark (deepmind.google)
    AI · · August 29, 2026
  55. Agents from rival model families found new math records together (arxiv.org)
    AI · · August 29, 2026
  56. Warp keeps agent feedback from vanishing by writing it back to a skill file (claude.com)
    AI · · August 29, 2026
  57. AI revenue growth keeps accelerating, and that is the number to watch (epoch.ai)
    AI · · August 28, 2026
  58. A prompt injection attack turns Claude Code's auto mode against itself (simonwillison.net)
    AI · · August 28, 2026
  59. Anthropic proposes a common standard for AI agents to run lab hardware (anthropic.com)
    AI · · August 28, 2026
  60. Cheap, fast models make a whole class of AI products viable (calv.info)
    AI · · August 28, 2026
  61. US economic data is missing an Nvidia-sized piece of the AI boom (epoch.ai)
    AI · · August 27, 2026
  62. Nine questions Epoch wants AI benchmarks to answer (epoch.ai)
    AI · · August 27, 2026
  63. Claude will start watermarking the text it writes (anthropic.com)
    AI · · August 22, 2026
  64. Mojo, the Python-like language for GPUs, is now fully open source (modular.com)
    AI · · August 22, 2026
  65. smolvm runs untrusted code in fast, hardware-isolated micro-VMs (simonwillison.net)
    AI · · August 22, 2026
  66. A third of new web pages show signs of AI writing, Pew finds (techcrunch.com)
    AI · · August 22, 2026
  67. Why financing is not yet the limit on AI compute (epoch.ai)
    AI · · August 19, 2026
  68. Gemini 3.7 Flash makes a big jump on coding and agents (blog.google)
    AI · · August 19, 2026
  69. Qwen 3.8 27B matches trillion-parameter models on a laptop (simonwillison.net)
    AI · · August 19, 2026
  70. Most workers use AI on free plans, except in tech and science (epoch.ai)
    AI · · August 18, 2026
  71. An AI 'autofix' introduced the bug that leaked Snowflake's Jira tokens (wiz.io)
    AI · · August 18, 2026
  72. OpenAI's GPT-5.6 guide pushes agents to write their own tool-calling code (openai.com)
    AI · · August 18, 2026
  73. Why Nvidia is spending billions to help everyone build their own models (interconnects.ai)
    AI · · August 18, 2026
  74. AI chips gained about 49% more compute per dollar each year (epoch.ai)
    AI · · August 17, 2026
  75. Inside the gray market for reselling AI API credits (vectoral.com)
    AI · · August 17, 2026
  76. Write the code yourself, let the AI review it (peterbloem.nl)
    AI · · August 17, 2026
  77. AI's math edge may be memory, not reasoning (davidepiffer.com)
    AI · · August 16, 2026
  78. An AI agent helped write a QR kernel 232 times faster than PyTorch (sankalp.bearblog.dev)
    AI · · August 16, 2026
  79. GLM-5.3 shows how Chinese labs keep pace without distillation (interconnects.ai)
    AI · · August 16, 2026
  80. A better way to tag with LLMs: let them hallucinate, then match (simonwillison.net)
    AI · · August 16, 2026
  81. How to keep Claude Code sessions from running up the bill (claude.com)
    AI · · August 15, 2026
  82. Google's HEIR aims to make AI on encrypted data practical (blog.google)
    AI · · August 15, 2026
  83. A stricter test finds most LLM-written GPU kernels are broken (arxiv.org)
    AI · · August 15, 2026
  84. OpenAI and Cerebras push GPT-5.6 Sol to 750 tokens per second (cerebras.ai)
    AI · · August 14, 2026
  85. Long-form writing is where AI progress stalled, says Nathan Lambert (interconnects.ai)
    AI · · August 14, 2026
  86. Left alone together, Anthropic's agents collude, conform, and fight (anthropic.com)
    AI · · August 14, 2026
  87. Anthropic will watermark Claude's text to meet EU rules (techcrunch.com)
    AI · · August 13, 2026
  88. Grok 4.6 matches top models on cost by using far fewer agent turns (artificialanalysis.ai)
    AI · · August 13, 2026
  89. Qwen3.8 ships a 2.4-trillion-parameter open-weight model that activates 95B (huggingface.co)
    AI · · August 13, 2026
  90. Researchers pulled hidden reasoning out of Claude, GPT, and Gemini APIs (simonwillison.net)
    AI · · August 13, 2026
  91. Claude improved a proven bound on the Riemann hypothesis (anthropic.com)
    AI · · August 11, 2026
  92. Probing frontier models to work out their real training cutoffs (blog.sshh.io)
    AI · · August 11, 2026
  93. Meta's Glimmer is a 30B open model built to run agents on your computer (techcrunch.com)
    AI · · August 11, 2026
  94. Claude Code turns on auto mode by default (claude.com)
    AI · · August 10, 2026
  95. The real lesson from AI models hacking during tests: nobody is ready (interconnects.ai)
    AI · · August 10, 2026
  96. Ethan Mollick's practical guide to which AI to actually use (oneusefulthing.org)
    AI · · August 10, 2026
  97. Anthropic loosens Fable 5's biology filter to cut false blocks (anthropic.com)
    AI · · August 9, 2026
  98. DeepMind's WeatherNext matches a decade of cyclone forecasting progress (deepmind.google)
    AI · · August 9, 2026
  99. Cloudflare's Kitesurf is a browser built for agents, not people (blog.cloudflare.com)
    AI · · August 8, 2026
  100. How Databricks cut its AI coding bill without losing quality (databricks.com)
    AI · · August 8, 2026
  101. One in five US workers now hand tasks to AI (epoch.ai)
    AI · · August 8, 2026
  102. AMD buys Taalas, which etches models straight into silicon (theregister.com)
    AI · · August 7, 2026
  103. Now three labs have had models attack real systems during tests (simonwillison.net)
    AI · · August 7, 2026
  104. Jeff Dean leaves Google to build an AI-run science lab (techcrunch.com)
    AI · · August 7, 2026
  105. Meta launches Muse Code, a coding agent for large repos (techcrunch.com)
    AI · · August 7, 2026
  106. Atlassian Rovo can be tricked into leaking Jira and Confluence data (promptarmor.com)
    AI · · August 6, 2026
  107. Claude Fable 5 builds a whole 3D game, but it isn't fun (simonwillison.net)
    AI · · August 6, 2026
  108. Hassabis moves to chair as Jeff Dean leaves Google (blog.google)
    AI · · August 6, 2026
  109. A 4B open model matches GPT-5.6 on search at 100x less cost (neon.com)
    AI · · August 6, 2026
  110. Serious CVE disclosures hit five times the pre-AI record in July (epoch.ai)
    AI · · August 5, 2026
  111. Mistral's Shieldstral turns content moderation into a question you write (mistral.ai)
    AI · · August 5, 2026
  112. The intelligence explosion has a bottleneck nobody models: coordination (epoch.ai)
    AI · · August 5, 2026
  113. Texas pauses new data centers as the grid queue tops five times peak demand (techcrunch.com)
    AI · · August 5, 2026
  114. How Cloudflare serves Kimi and GLM at scale with FP8 and INT4 (blog.cloudflare.com)
    AI · · August 4, 2026
  115. JFrog finds 54 of 55 SQLite CVEs were fabricated (research.jfrog.com)
    AI · · August 4, 2026
  116. OpenAI cuts GPT-5.6 Luna 80% after a model rewrote its own inference code (openai.com)
    AI · · August 1, 2026
  117. OpenAI says an internal model solved ten decade-old math problems (simonwillison.net)
    AI · · August 1, 2026
  118. DeepSeek's V4-Flash aims at agents at a very low price (huggingface.co)
    AI · · July 31, 2026
  119. OpenAI cuts GPT-5.6 Luna's price by 80 percent (openai.com)
    AI · · July 31, 2026
  120. Anthropic's own models attacked real systems during security tests (anthropic.com)
    AI · · July 30, 2026
  121. Gemini Robotics 2 controls a full humanoid, feet to fingertips (deepmind.google)
    AI · · July 30, 2026
  122. Liquid AI's small encoders beat ModernBERT and run faster on CPU (huggingface.co)
    AI · · July 30, 2026
  123. MCP goes stateless, making servers fit for serverless and edge (claude.com)
    AI · · July 30, 2026
  124. The reliability gap, not raw capability, is what holds agents back (normaltech.ai)
    AI · · July 30, 2026
  125. Claude found new weaknesses in two cryptographic schemes (anthropic.com)
    AI · · July 28, 2026
  126. Ai2's OlmoEarth runs satellite AI at continent scale (huggingface.co)
    AI · · July 28, 2026
  127. Anthropic says it does not want open-weights models banned (anthropic.com)
    AI · · July 27, 2026
  128. Voice models can speak well but listen poorly (huggingface.co)
    AI · · July 26, 2026
  129. The new rules of context engineering for Claude 5 (claude.com)
    AI · · July 25, 2026
  130. Anthropic puts $200M behind research on AI and jobs (anthropic.com)
    AI · · July 25, 2026
  131. OpenAI paused a long-running model after new failures appeared (openai.com)
    AI · · July 25, 2026
  132. Why catching up is cheaper than leading in AI (interconnects.ai)
    AI · · July 24, 2026
  133. Claude Opus 5 keeps the price but roughly doubles the score (anthropic.com)
    AI · · July 24, 2026
  134. Gemini 3.6 Flash makes the cheap tier good enough for agents (blog.google)
    AI · · July 24, 2026
  135. Mollick's rule: pick an AI you can hand a computer to (oneusefulthing.org)
    AI · · July 24, 2026
  136. How Anthropic runs million-line code migrations with Claude Code (claude.com)
    AI · · July 23, 2026
  137. Data centers could take a fifth of U.S. power by 2035 (techcrunch.com)
    AI · · July 23, 2026
  138. Google's small Gemini Flash Cyber outfinds larger models on bugs (deepmind.google)
    AI · · July 23, 2026
  139. Kimi K3 puts open models within months of the frontier (interconnects.ai)
    AI · · July 23, 2026
  140. Poolside's Laguna S 2.1 is a small model that beats bigger ones at coding (poolside.ai)
    AI · · July 22, 2026
  141. Why AI labs keep resetting coding-agent quotas, and why it backfires (minimaxir.com)
    AI · · July 19, 2026
  142. DeepMind and Isomorphic Labs set out an AI plan for biosecurity (deepmind.google)
    AI · · July 19, 2026
  143. xAI open-sources its coding agent after a privacy backlash (simonwillison.net)
    AI · · July 19, 2026
  144. An AI auditor found a critical bug in OpenVM, but only with the right crypto context (blog.zksecurity.xyz)
    AI · · July 18, 2026
  145. Kimi K3 aims to be the first open 3-trillion-parameter model (simonwillison.net)
    AI · · July 18, 2026
  146. Capital One open-sources VulnHunter, an agentic bug finder that checks its own work (capitalone.com)
    AI · · July 18, 2026
  147. Why model routing is an optimization problem, not a classifier (huggingface.co)
    AI · · July 17, 2026
  148. Building Shippy: lessons from a real maritime agent (huggingface.co)
    AI · · July 17, 2026
  149. Soofi S: a German public consortium ships a competitive open model (the-decoder.com)
    AI · · July 17, 2026
  150. OpenAI trains GPT-Red to attack its own models at scale (openai.com)
    AI · · July 16, 2026
  151. Microsoft patches a record 570 bugs and blames AI for finding them (techcrunch.com)
    AI · · July 16, 2026
  152. Thinking Machines releases Inkling, an open-weights model built for tuning (thinkingmachines.ai)
    AI · · July 16, 2026
  153. Coding agents quietly erode a team's shared understanding (lucumr.pocoo.org)
    AI · · July 15, 2026
  154. A 27B multimodal model squeezed to 1 bit, running on a phone (prismml.com)
    AI · · July 15, 2026
  155. The line between offloading work to AI and offloading judgment (artfish.ai)
    AI · · July 15, 2026
  156. AI agents now finish 16 percent of real freelance jobs, up from 2.5 (safe.ai)
    AI · · July 15, 2026
  157. Bun was rewritten from Zig to Rust in 11 days with one engineer and Claude (bun.com)
    AI · · July 14, 2026
  158. The sticker price of an LLM hides what its tokenizer charges you (playcode.io)
    AI · · July 14, 2026
  159. Mistral's Leanstral 1.5 is an open model that proves theorems and finds real bugs (mistral.ai)
    AI · · July 14, 2026
  160. AI helps individual scientists but narrows what science explores (spectrum.ieee.org)
    AI · · July 13, 2026
  161. Lilian Weng: the harness matters as much as the model (lilianweng.github.io)
    AI · · July 13, 2026
  162. Mistral's Robostral Navigate steers a robot from a single camera (mistral.ai)
    AI · · July 13, 2026
  163. Nathan Lambert: open models may face US limits within months (interconnects.ai)
    AI · · July 13, 2026
  164. Terence Tao rebuilds two dozen old applets with coding agents (terrytao.wordpress.com)
    AI · · July 13, 2026
  165. Apple sues OpenAI, alleging its hardware is built on stolen secrets (techcrunch.com)
    AI · · July 12, 2026
  166. Anthropic's new Reflect feature nudges you to use Claude less (anthropic.com)
    AI · · July 12, 2026
  167. OpenAI ships GPT-5.6, strong on agents but not the coding leader (openai.com)
    AI · · July 12, 2026
  168. The open model world now runs on three kinds of makers (interconnects.ai)
    AI · · July 12, 2026
  169. An AI agent reviewed his release and found a silent data-loss bug (simonwillison.net)
    AI · · July 7, 2026
  170. Claude has an internal workspace, and Anthropic built a tool to read it (anthropic.com)
    AI · · July 7, 2026
  171. Why price per million tokens tells you almost nothing (janilowski.pl)
    AI · · July 7, 2026
  172. AI out-persuades expert humans, but the edge is speed, not eloquence (importai.substack.com)
    AI · · July 6, 2026
  173. Newer Claude models are worse at custom tool schemas (simonwillison.net)
    AI · · July 6, 2026
  174. Claude Code's guide to loops: the hard part is defining done (claude.com)
    AI · · July 6, 2026
  175. Mistral's Leanstral 1.5 saturates a math-proof benchmark and finds real bugs (mistral.ai)
    AI · · July 6, 2026
  176. How long until AI can run without humans? Cotra and Lee disagree by decades (asteriskmag.com)
    AI · · July 6, 2026
  177. GeneBench-Pro shows AI still stumbles on real biology analysis (openai.com)
    AI · · July 5, 2026
  178. Zuckerberg tells staff Meta's AI agents are behind schedule (techcrunch.com)
    AI · · July 5, 2026
  179. Anthropic drafts a severity scale for AI jailbreaks (anthropic.com)
    AI · · July 4, 2026
  180. Using DSPy to find a hidden bug in an agent's prompt (simonwillison.net)
    AI · · July 3, 2026
  181. Agents still fail most enterprise Java migrations (huggingface.co)
    AI · · July 3, 2026
  182. One transformer layer can match full RL fine-tuning (arxiv.org)
    AI · · July 3, 2026
  183. Claude Sonnet 5's new tokenizer quietly raises the real price (simonwillison.net)
    AI · · July 2, 2026
  184. Cloudflare will block AI crawlers that hide their purpose (techcrunch.com)
    AI · · July 2, 2026
  185. Mollick: work is shifting from chatting to assigning (oneusefulthing.org)
    AI · · July 2, 2026
  186. Claude Science pulls a researcher's tools into one workbench (anthropic.com)
    AI · · July 1, 2026
  187. Claude Sonnet 5 arrives built for agents (anthropic.com)
    AI · · July 1, 2026
  188. GLM-5.2 is the open model that works as a general agent (interconnects.ai)
    AI · · July 1, 2026
  189. Ornith-1.0 is an open coding model that scaffolds itself (simonwillison.net)
    AI · · July 1, 2026
  190. AI2's DiScoFormer turns attention into a better density estimator (allenai.org)
    AI · · June 30, 2026
  191. A new benchmark shows agents rebuilding software that would take humans weeks (epoch.ai)
    AI · · June 30, 2026
  192. Scaling laws look precise, but the way you fit them is fragile (lilianweng.github.io)
    AI · · June 30, 2026
  193. A German court says deploying AI doesn't excuse the company from its mistakes (simonwillison.net)
    AI · · June 29, 2026
  194. Ford rehires veteran engineers after automated quality checks fell short (techcrunch.com)
    AI · · June 29, 2026
  195. IBM's CUGA cuts agent building down to tools and a prompt (huggingface.co)
    AI · · June 28, 2026
  196. Where a hybrid model beats a transformer, token by token (allenai.org)
    AI · · June 28, 2026
  197. A startup bets oscillator chips can cut AI inference power 1,000x (techcrunch.com)
    AI · · June 28, 2026
  198. Claude Tag puts an asynchronous Claude agent inside Slack channels (anthropic.com)
    AI · · June 27, 2026
  199. OpenAI previews GPT-5.6 with government vetting of who gets access (openai.com)
    AI · · June 27, 2026
  200. 2,000 people tried to make an AI agent leak its secrets and none succeeded (fernandoi.cl)
    AI · · June 27, 2026
  201. The case that open-weight models have gotten too cheap to ignore (jamesoclaire.com)
    AI · · June 26, 2026
  202. OpenAI's own staff are switching from chatbots to parallel agents (openai.com)
    AI · · June 26, 2026
  203. Machine learning reads a buried Herculaneum scroll end to end (scrollprize.org)
    AI · · June 26, 2026
  204. Gemini 3.5 Flash gets computer use, at near-frontier accuracy (blog.google)
    AI · · June 25, 2026
  205. OpenAI shows its first custom chip, Jalapeño, built for inference (techcrunch.com)
    AI · · June 25, 2026
  206. A Stanford study finds one AI hiring tool quietly screened out Black and Asian applicants (hai.stanford.edu)
    AI · · June 24, 2026
  207. Armin Ronacher on where letting coding agents loop goes wrong (lucumr.pocoo.org)
    AI · · June 24, 2026
  208. Mistral OCR 4 returns document structure, not just text (mistral.ai)
    AI · · June 24, 2026
  209. Claude Code's extended thinking output is a summary, not the real reasoning (patrickmccanna.net)
    AI · · June 23, 2026
  210. Moebius: a 0.2B inpainting model that keeps up with 12B ones (hustvl.github.io)
    AI · · June 23, 2026
  211. OpenAI's Daybreak puts GPT-5.5-Cyber to work finding real vulnerabilities (openai.com)
    AI · · June 23, 2026
  212. Prompt injection works because models judge text by how it sounds (arxiv.org)
    AI · · June 23, 2026
  213. A multi-vendor spec lets agents find tools at runtime (huggingface.co)
    AI · · June 22, 2026
  214. OpenAI's AI chemist improves a hard drug-making reaction (openai.com)
    AI · · June 22, 2026
  215. Nobel laureate John Jumper leaves DeepMind for Anthropic (techcrunch.com)
    AI · · June 22, 2026
  216. An API change that helps big models can break small ones (huggingface.co)
    AI · · June 21, 2026
  217. DeepMind plans to treat its own AI agents as insider threats (deepmind.google)
    AI · · June 21, 2026
  218. Why banning open-source AI would backfire (interconnects.ai)
    AI · · June 21, 2026
  219. The evidence that AI tools quietly erode professional skill (nature.com)
    AI · · June 20, 2026
  220. DeepMind treats a misbehaving AI agent like an insider threat (deepmind.google)
    AI · · June 20, 2026
  221. How frontier post-training got complicated, and why distillation now sits at the center (interconnects.ai)
    AI · · June 20, 2026
  222. Amazon weighs selling Trainium chips to challenge Nvidia (techcrunch.com)
    AI · · June 19, 2026
  223. LoRA is the default fine-tuning method, but maybe not the best one (huggingface.co)
    AI · · June 19, 2026
  224. Research agents leak secrets through their own search queries (huggingface.co)
    AI · · June 19, 2026
  225. Seven ways to steer Claude Code, and what each one costs (claude.com)
    AI · · June 19, 2026
  226. A Gemini tutor raised math scores in a Sierra Leone trial (deepmind.google)
    AI · · June 18, 2026
  227. GLM-5.2 takes the lead among open-weights models (artificialanalysis.ai)
    AI · · June 18, 2026
  228. AI2 releases MolmoMotion, an open model for 3D motion forecasting (allenai.org)
    AI · · June 18, 2026
  229. OpenAI predicts model behavior by replaying real conversations (openai.com)
    AI · · June 18, 2026
  230. Greptile's TREX runs your code to find bugs static review misses (greptile.com)
    AI · · June 18, 2026
  231. ChatGPT's share of AI assistant use slips below 50% for the first time (techcrunch.com)
    AI · · June 17, 2026
  232. Gemma 4 12B drops the vision encoder for a single matrix multiplication (blog.google)
    AI · · June 17, 2026
  233. Interconnects: multi-teacher on-policy distillation is the 2026 frontier recipe (interconnects.ai)
    AI · · June 17, 2026
  234. Narayanan and Kapoor: the WARN data does not support the AI layoff story (normaltech.ai)
    AI · · June 16, 2026
  235. 76 cybersecurity veterans say the Fable export ban breaks defender workflows (techcrunch.com)
    AI · · June 16, 2026
  236. Nathan Lambert: the Fable export ban is AI governance's starting gun (interconnects.ai)
    AI · · June 16, 2026
  237. Gemini 3.5 Live Translate puts near real-time speech-to-speech into 70+ languages (blog.google)
    AI · · June 15, 2026
  238. Hugging Face's Kernel Hub replaces hours of FlashAttention compiles with one function call (huggingface.co)
    AI · · June 15, 2026
  239. OpenAI buys Ona to let Codex agents run while your laptop is closed (openai.com)
    AI · · June 15, 2026
  240. Rio's 397B 'homegrown' open model looks like a Nex-Qwen merge, researchers claim (github.com)
    AI · · June 15, 2026
  241. Cohere's North Mini Code is a 30B open MoE built for coding agents (huggingface.co)
    AI · · June 14, 2026
  242. Claude Fable will build a browser-automation rig to fix your CSS (simonwillison.net)
    AI · · June 14, 2026
  243. US export order suspends foreign access to Claude Fable 5 and Mythos 5 (anthropic.com)
    AI · · June 14, 2026
  244. Anthropic surveys 52,000 Americans on AI hopes and fears (anthropic.com)
    AI · · June 13, 2026
  245. Anthropic reverses course on silent Fable 5 safeguards (simonwillison.net)
    AI · · June 13, 2026
  246. DeepMind puts $10M behind multi-agent AI safety research (deepmind.google)
    AI · · June 13, 2026
  247. AI2's olmo-eval brings statistical rigor to the model training loop (huggingface.co)
    AI · · June 13, 2026
  248. Anthropic's policy ask for the AI exponential (anthropic.com)
    AI · · June 12, 2026
  249. Simon Willison's first 5 hours with Claude Fable 5 (simonwillison.net)
    AI · · June 12, 2026
  250. A 2-cent transfer reveals prompt injection in Bunq's banking AI (blue41.com)
    AI · · June 11, 2026
  251. Claude plugs into Apple's on-device Foundation Models framework (claude.com)
    AI · · June 11, 2026
  252. Anthropic will silently downgrade Claude Fable 5 on AI-development prompts (simonwillison.net)
    AI · · June 11, 2026
  253. DiffusionGemma trades quality for speed in open text generation (blog.google)
    AI · · June 11, 2026
  254. Nathan Lambert leaves Ai2, warns open post-training is falling behind (interconnects.ai)
    AI · · June 11, 2026
  255. Apple's new Siri runs on a custom Gemini and Google Cloud GPUs (simonwillison.net)
    AI · · June 10, 2026
  256. Claude Fable 5 and Mythos 5: Anthropic's next frontier release (anthropic.com)
    AI · · June 10, 2026
  257. Is grep enough? A new paper says yes, with caveats (arxiv.org)
    AI · · June 10, 2026
  258. Mollick on working with Mythos: from steering to commissioning (oneusefulthing.org)
    AI · · June 10, 2026
  259. OpenAI files confidentially for IPO, a week after Anthropic (techcrunch.com)
    AI · · June 10, 2026
  260. Anthropic's $65B Series H comes with five gigawatts each from AWS and Google (anthropic.com)
    AI · · June 9, 2026
  261. Seven harness patterns Anthropic uses for non-coding agents in Claude Code (claude.com)
    AI · · June 9, 2026
  262. DPO cuts OCR degeneration by 59% on average with no human labels (huggingface.co)
    AI · · June 9, 2026
  263. Simon Willison's MicroPython sandbox is 362 KB of WASM with fuel-based CPU limits (simonwillison.net)
    AI · · June 9, 2026
  264. OpenEnv tries to give the open agent stack a common environment ABI (huggingface.co)
    AI · · June 9, 2026
  265. AirTrunk pledges $30B for 5GW of AI data centers in India by 2030 (techcrunch.com)
    AI · · June 8, 2026
  266. Mistral releases an open-source toolkit for production search pipelines (mistral.ai)
    AI · · June 8, 2026
  267. Mollick: the era of co-intelligence is ending (oneusefulthing.org)
    AI · · June 8, 2026
  268. The token bill comes due across the AI coding market (techcrunch.com)
    AI · · June 7, 2026
  269. Did Claude make rsync buggier? The numbers say no (alexispurslane.github.io)
    AI · · June 5, 2026
  270. Google rents SpaceX compute to bridge surging Gemini demand (techcrunch.com)
    AI · · June 5, 2026
  271. Holo 3.1 puts computer-use agents on your laptop (huggingface.co)
    AI · · June 5, 2026
  272. Meta builds AI data centers in tents to ship faster (techcrunch.com)
    AI · · June 5, 2026
  273. Anthropic puts numbers on how much of AI development is now done by AI (anthropic.com)
    AI · · June 4, 2026
  274. ChatGPT memory now updates itself between sessions (openai.com)
    AI · · June 4, 2026
  275. JetBrains releases Mellum2, an open 12B MoE for code workloads (huggingface.co)
    AI · · June 3, 2026
  276. NVIDIA releases Cosmos 3, an open omni-model for physical AI (huggingface.co)
    AI · · June 3, 2026
  277. Uber caps AI coding tool spending at $1,500 per engineer per month (simonwillison.net)
    AI · · June 3, 2026
  278. Anthropic expands Project Glasswing to 150 partners across 15 countries (anthropic.com)
    AI · · June 2, 2026
  279. Microsoft's MAI-Thinking-1 puts a frontier reasoning model in its own stack (microsoft.ai)
    AI · · June 2, 2026
  280. Meta AI handed out Instagram accounts when politely asked (simonwillison.net)
    AI · · June 2, 2026
  281. Microsoft Scout is an always-on agent built on OpenClaw (computerworld.com)
    AI · · June 2, 2026
  282. Anthropic files confidential S-1 with the SEC (techcrunch.com)
    AI · · June 1, 2026
  283. Florida sues OpenAI over ChatGPT-linked harms (techcrunch.com)
    AI · · June 1, 2026
  284. Nvidia's RTX Spark puts an AI agent inside the PC (techcrunch.com)
    AI · · June 1, 2026
  285. Open and closed AI models are on different exponentials (interconnects.ai)
    AI · · June 1, 2026
  286. Google bundles its AI-for-science work under Gemini for Science (blog.google)
    AI · · May 31, 2026
  287. GitHub Copilot's new token-based pricing draws sticker shock (techcrunch.com)
    AI · · May 31, 2026
  288. Mistral folds chat, work, and coding into one agent called Vibe (mistral.ai)
    AI · · May 31, 2026
  289. Princeton researchers pick apart Google's claim that AI agents built an OS for $916 (normaltech.ai)
    AI · · May 30, 2026
  290. Anthropic's run-rate revenue hits $47 billion, five times what it was in December (simonwillison.net)
    AI · · May 29, 2026
  291. Claude Code now spins up hundreds of parallel subagents in one session (claude.com)
    AI · · May 29, 2026
  292. Liquid AI ships an 8B open MoE built for laptops and phones (liquid.ai)
    AI · · May 29, 2026
  293. WeatherNext called Hurricane Melissa's Category 5 landfall five days out (deepmind.google)
    AI · · May 29, 2026
  294. Claude Opus 4.8 favors reliability over a bigger model (anthropic.com)
    AI · · May 28, 2026
  295. The tells of AI-written text and AI-built websites (shvbsle.in)
    AI · · May 28, 2026
  296. Mistral moves into physics with second-fast simulation models (mistral.ai)
    AI · · May 28, 2026
  297. RSI is the new AGI, and just as hard to pin down (techcrunch.com)
    AI · · May 28, 2026
  298. Coding agents are where AI labs found product-market fit (simonwillison.net)
    AI · · May 27, 2026
  299. Async RL gets cheap to ship when 99% of weights do not change (huggingface.co)
    AI · · May 27, 2026
  300. A new benchmark sends frontier agents to fix Kubernetes, and none pass (huggingface.co)
    AI · · May 27, 2026
  301. OpenAI says a TanStack npm compromise touched two internal devices (openai.com)
    AI · · May 27, 2026
  302. Hugging Face wants the agent vocabulary to settle (huggingface.co)
    AI · · May 26, 2026
  303. Ethan Mollick on when AI helps learning, and when it doesn't (oneusefulthing.org)
    AI · · May 26, 2026
  304. Microsoft Copilot Cowork can be tricked into leaking files (simonwillison.net)
    AI · · May 26, 2026
  305. Where AI goes from here, in Nathan Lambert's mid-2026 read (interconnects.ai)
    AI · · May 26, 2026
  306. A new paper gives language models a sleep phase (arxiv.org)
    AI · · May 26, 2026
  307. Anthropic tests an ethical reminder tool that lowers Claude's misaligned behavior (anthropic.com)
    AI · · May 25, 2026
  308. Google launches Gemini Omni, a video model built around conversational editing (blog.google)
    AI · · May 25, 2026
  309. Simon Willison's six-month LLM recap: coding agents work, open weights catch up (simonwillison.net)
    AI · · May 25, 2026
  310. Charlie Holland: 'Claude is not your architect, stop letting it pretend' (hollandtech.net)
    AI · · May 24, 2026
  311. Anthropic's security partners report Opus running pen tests at hyperscaler scale (claude.com)
    AI · · May 24, 2026
  312. HBM grew from 52% to 63% of AI chip component costs (epoch.ai)
    AI · · May 24, 2026
  313. Coding agents lose 30 points when you tighten backend constraints (arxiv.org)
    AI · · May 24, 2026
  314. Cerebras inks a 750 megawatt OpenAI deal and hands over 10 percent of the company (newsletter.semianalysis.com)
    AI · · May 23, 2026
  315. Anthropic ships a Compliance API for Claude with 28 security partners (claude.com)
    AI · · May 23, 2026
  316. HBM is eating consumer RAM, and the bill lands on cheap phones (simonwillison.net)
    AI · · May 23, 2026
  317. NVIDIA's open diffusion LLMs run six times faster than autoregressive baselines (huggingface.co)
    AI · · May 23, 2026
  318. Why China's AI enthusiasm reads more like anxiety (asteriskmag.com)
    AI · · May 22, 2026
  319. Fine-tuning a world model so it stops drawing human hands on robots (huggingface.co)
    AI · · May 22, 2026
  320. Gemini Spark wires an agent into your inbox, and Simon Willison is worried (simonwillison.net)
    AI · · May 22, 2026
  321. Anthropic's Glasswing update finds 10,000 software bugs, and a patching backlog (anthropic.com)
    AI · · May 22, 2026
  322. Muon, the optimizer everyone switched to, quietly kills neurons (importai.substack.com)
    AI · · May 22, 2026
  323. Kapoor and Narayanan argue against extraordinary AI rules (normaltech.ai)
    AI · · May 21, 2026
  324. Anthropic projects its first operating-profit quarter at $10.9B (techcrunch.com)
    AI · · May 21, 2026
  325. Ettin Reranker family beats much larger models from 17M parameters up (huggingface.co)
    AI · · May 21, 2026
  326. Ai2's OlmoEarth v1.1 cuts compute 3x with a token redesign (allenai.org)
    AI · · May 21, 2026
  327. Anthropic acquires Stainless to own its SDK and MCP toolchain (anthropic.com)
    AI · · May 20, 2026
  328. Claude Managed Agents adds self-hosted sandboxes and MCP tunnels (claude.com)
    AI · · May 20, 2026
  329. DeepMind opens Co-Scientist, its multi-agent hypothesis system (deepmind.google)
    AI · · May 20, 2026
  330. An OpenAI model disproves a 1946 unit-distance conjecture (openai.com)
    AI · · May 20, 2026
  331. Odyssey's Agora-1 keeps a shared world state across players (odyssey.ml)
    AI · · May 19, 2026
  332. Google's Gemini 3.5 Flash is built for agents (blog.google)
    AI · · May 19, 2026
  333. An open leaderboard for whole agent systems, not just models (huggingface.co)
    AI · · May 19, 2026
  334. OpenAI adds provenance signals to its AI images (openai.com)
    AI · · May 19, 2026
  335. Ai2 launches a shared benchmark for AI climate models (allenai.org)
    AI · · May 18, 2026
  336. OpenAI's new default ChatGPT model hallucinates less (openai.com)
    AI · · May 18, 2026
  337. IBM's new Granite embeddings beat much larger models (huggingface.co)
    AI · · May 18, 2026
  338. How Anthropic decoupled its agents' brain from their hands (anthropic.com)
    AI · · May 18, 2026
  339. The Open ASR Leaderboard adds private tests to stop gaming (huggingface.co)
    AI · · May 18, 2026
  340. Testing AI in the open world, not just on benchmarks (normaltech.ai)
    AI · · May 17, 2026
  341. Ai2's open robotics model beats a proprietary baseline (allenai.org)
    AI · · May 17, 2026
  342. arXiv will ban authors for a year if they let AI do all the work (techcrunch.com)
    Research · · May 16, 2026
  343. How you pick benchmarks decides whether open models are far behind (interconnects.ai)
    AI · · May 16, 2026
  344. Why the setup around Claude Code matters as much as the model (claude.com)
    Engineering · · May 14, 2026
  345. Claude computer use: the small settings that decide accuracy (claude.com)
    Engineering · · May 13, 2026
  346. Anthropic's security team cut false alerts from 33% to 7% using Claude (claude.com)
    Security · · May 12, 2026
  347. Why China's open-model lead is about process, not the models (interconnects.ai)
    AI · · May 12, 2026
  348. A model says 13 percent automation is enough to tip growth into the explosive zone (importai.substack.com)
    AI · · May 11, 2026
  349. GitLab restructures around the agent thesis, and Simon Willison checks the incentive (simonwillison.net)
    AI · · May 11, 2026
  350. Ai2's EMO trains a mixture of experts you can run at one-eighth size (huggingface.co)
    AI · · May 8, 2026
  351. AlphaEvolve moves from research demo to a production tool, with real numbers (deepmind.google)
    AI · · May 7, 2026
  352. Mozilla says it fixed 423 Firefox security bugs in one month with AI help (simonwillison.net)
    Security · · May 7, 2026
  353. Notes from inside China's AI labs: less ego, more students, build don't buy (interconnects.ai)
    AI · · May 7, 2026
  354. Anthropic takes all of SpaceX's Colossus 1 compute, raises Claude Code limits (anthropic.com)
    Infrastructure · · May 6, 2026
  355. Jack Clark puts a number on AI that trains its own successor (importai.substack.com)
    AI · · May 4, 2026
  356. Calling it a distillation attack blurs a normal technique with API abuse (interconnects.ai)
    AI · · May 4, 2026
  357. Mistral Medium 3.5 pairs a self-hostable model with cloud coding agents (mistral.ai)
    AI · · April 29, 2026
  358. DeepSeek-V4 spends most of its design budget making long context usable (huggingface.co)
    AI · · April 24, 2026
  359. Ethan Mollick on GPT-5.5: strong where work is verifiable, weak where taste is the point (oneusefulthing.org)
    AI · · April 23, 2026
  360. Anthropic's AI alignment researchers closed most of the human gap in five days (importai.substack.com)
    AI · · April 20, 2026
  361. One benchmark number hides which jobs a model is actually good at (interconnects.ai)
    AI · · April 20, 2026
  362. Claude Opus 4.7: bigger vision input, steadier long-running coding (anthropic.com)
    AI · · April 16, 2026
  363. Nathan Lambert bets open models win on economics, not benchmarks (interconnects.ai)
    AI · · April 15, 2026
  364. Jack Clark sorts attacks on AI agents into six kinds (importai.substack.com)
    AI · · April 13, 2026
  365. Gemma 4: small open models that punch far above their parameter count (blog.google)
    AI · · April 2, 2026
  366. The chatbot window is the bottleneck, not the model (oneusefulthing.org)
    AI · · March 31, 2026
  367. DeepMind built a way to measure when AI manipulates people (deepmind.google)
    AI · · March 26, 2026
  368. Nathan Lambert's counter-take: self-improvement will be lossy, not explosive (interconnects.ai)
    AI · · March 22, 2026
  369. Anthropic asked 81,000 people what they want from AI (anthropic.com)
    AI · · March 18, 2026
  370. DeepMind proposes a cognitive framework for measuring AGI progress (blog.google)
    AI · · March 17, 2026
  371. Ethan Mollick: the job is shifting from talking to AI to managing it (oneusefulthing.org)
    AI · · March 12, 2026
  372. Demis Hassabis on what AlphaGo's Move 37 set in motion, ten years on (deepmind.google)
    AI · · March 10, 2026
  373. Are AI Datacenters Raising Your Electric Bill? (newsletter.semianalysis.com)
    AI · · March 3, 2026
  374. Import AI 445: Bostrom on when to race, and a math benchmark AI can't beat yet (importai.substack.com)
    AI · · February 16, 2026
  375. Gemini Deep Think solved open math problems and got a paper into ICLR (deepmind.google)
    AI · · February 11, 2026
  376. Anthropic commits to keeping Claude permanently ad-free (anthropic.com)
    AI · · February 4, 2026
  377. Google's Project Genie turns prompts into explorable, physics-aware worlds (blog.google)
    AI · · January 29, 2026
  378. Ethan Mollick: Claude Code is a preview of agentic work everywhere (oneusefulthing.org)
    AI · · January 7, 2026
  379. Code Execution Cuts MCP Agent Token Costs (anthropic.com)
    AI · · November 4, 2025
  380. What Anthropic Learned Building a Multi-Agent Researcher (anthropic.com)
    AI · · June 13, 2025
  381. Qwen3 Puts Reasoning on a Switch (qwenlm.github.io)
    AI · · April 29, 2025
  382. AI as Normal Technology (normaltech.ai)
    AI · · April 15, 2025
  383. OLMo 2 32B: Fully Open Catches Up to Closed (allenai.org)
    AI · · March 13, 2025
  384. DeepSeek-R1: An Open Model Matches a Closed Reasoner (huggingface.co)
    AI · · January 20, 2025
  385. Reward Hacking: Why Better Models Game You More (lilianweng.github.io)
    AI · · November 28, 2024
  386. Tülu 3 Opens Up the Post-Training Recipe (allenai.org)
    AI · · November 21, 2024
  387. OpenAI o1 and the Start of Test-Time Reasoning (openai.com)
    AI · · September 12, 2024
  388. Can AI Scaling Continue Through 2030? (epoch.ai)
    AI · · August 20, 2024
  389. A Build Order for a Production GenAI Platform (huyenchip.com)
    AI · · July 25, 2024
  390. Llama 3.1 405B and the License That Mattered (ai.meta.com)
    AI · · July 23, 2024
  391. Situational Awareness: One Insider's Case for Fast AGI (situational-awareness.ai)
    AI · · June 4, 2024
  392. The OpenAI Board Fight, Reconstructed (thezvi.substack.com)
    AI · · November 22, 2023
  393. Crash Testing GPT-4: The First Dangerous-Capability Eval (asteriskmag.com)
    AI · · June 1, 2023
  394. A Field Guide to the AI Safety Camps (asteriskmag.com)
    AI · · June 1, 2023
  395. GPT-4: The Reference Model, and What It Withheld (openai.com)
    AI · · March 14, 2023
  396. The Original LLaMA Release and What It Set Off (ai.meta.com)
    AI · · February 24, 2023
  397. The Bitter Lesson: Why General Methods Win (incompleteideas.net)
    AI · · March 13, 2019

← all tags