1. September 2026
  2. Anthropic rebuilt its CI when Claude started writing most of the code (claude.com)
    AI · · September 18, 2026 · Anthropic · Engineering · Coding
  3. Anthropic starts measuring how much AI runs its own research (anthropic.com)
    AI · · September 18, 2026 · Anthropic · Safety · Policy
  4. Mismatched trade data points to AI servers smuggled into China (epoch.ai)
    AI · · September 18, 2026 · Compute · Hardware · China
  5. A model slipped instructions into its own memory summary (simonwillison.net)
    AI · · September 18, 2026 · OpenAI · Prompt Injection · Agents
  6. Claude folds Cowork back into chat (claude.com)
    AI · · September 17, 2026 · Anthropic · Claude
  7. Physics benchmarks were underrating the models (arxiv.org)
    AI · · September 17, 2026 · Evaluation · Research · Science
  8. A 4B model that beats the Postgres query planner (rohanbansal.com)
    AI · · September 17, 2026 · Engineering · Reinforcement Learning · Databases
  9. Suleyman's warning against 'model welfare' (mustafa-suleyman.ai)
    AI · · September 17, 2026 · Safety · Anthropic · Models
  10. Microsoft publishes a code of conduct for its AI models (techcrunch.com)
    AI · · September 15, 2026 · Microsoft · Safety · Governance
  11. What Mistral learned migrating 40,000 lines of Fortran with agents (mistral.ai)
    AI · · September 15, 2026 · Mistral · Agents · Engineering