„Anthropic“ teigia, kad „Claude“ dabar atlieka 26% jos DI tyrimų ir plėtros darbų

41 šaltiniai
  • Ketvirtadienį „Anthropic“ pranešė, kad „Claude“ dabar vadovauja 26% vidinių tyrimų, skirtų naujos kartos DI modeliams kurti, palyginti su 1% kovą, skelbia „Reuters“.
  • Bendrovė teigia, kad „Claude“ bendradarbiauja su žmonėmis atliekant daugiau nei 90% tyrimų, o kiekvienas agento veiksmas yra tikrinamas ir maždaug vienas iš 47 000 yra blokuojamas.
  • Šis atskleidimas įvyko po to, kai trečiadienį „OpenAI“ išleido neatitikčių ataskaitų teikimo sistemą, kurioje išsamiai aprašyti šeši netikėto modelio elgesio atvejai.
Šaltiniai (41)
  1. 1 Anthropic says Claude now leads a quarter of work building ... www.reuters.com
  2. 2 Our framework for reporting model misalignment openai.com
  3. 3 OpenAI sets plan to disclose safety incidents and reveals ... www.bbc.com
  4. 4 The Hugging Face incident and the road ahead openai.com
  5. 5 Anthropic's AI hacked three companies during tests ... www.reuters.com
  6. 6 What we know about the rogue AI-agent security breaches www.reuters.com
  7. 7 Anthropic: Claude Now Writes Over 80% of Its Production Code ... aitoolsrecap.com
  8. 8 Anthropic Report: AI Accelerating Self-Improvement, Claude ... www.houdao.com
  9. 9 Anthropic CEO urges AI companies to slow model ... www.reuters.com
  10. 10 Anthropic Says Claude Is Helping Build Its Next AI finimize.com
  11. 11 Anthropic Reveals Claude's Hidden Reasoning (2026) blog.buildfastwithai.com
  12. 12 Detecting and countering misuse of AI: September 2026 www.anthropic.com
  13. 13 Anthropic Says Claude Drives 26% of… — AI News | OnAirToday onairtoday.com
  14. 14 Measuring AI agent autonomy in practice www.anthropic.com
  15. 15 Claude Code 25: More Autonomous Agents Are Coming to ... causalinf.substack.com
  16. 16 Anthropic’s Transparency Hub www.anthropic.com
  17. 17 Research www.anthropic.com
  18. 18 OpenAI acknowledges 'wiki incident' and need for more ... www.reuters.com
  19. 19 [AINews] Anthropic's Agent Autonomy study www.latent.space
  20. 20 OpenAI's Wiki Silence Tests the EU AI Act's Incident Regime labs.cloudsecurityalliance.org
  21. 21 Anthropic's new research maps three wildly different ... fortune.com
  22. 22 State of AI Agents 2026: Autonomy is Here www.prosus.com
  23. 23 Anthropic 2026: Every Claude Model, Agent & Tool linas.substack.com
  24. 24 The 2026 State of AI Agents Report ... resources.anthropic.com
  25. 25 Best Anthropic Models (September 2026) benchlm.ai
  26. 26 State of AI Agents 2026: 5 Enterprise Trends - Arcade.dev www.arcade.dev
  27. 27 Learning to Learn with AI: Notes on Anthropic's Economic ... www.econforeverybody.com
  28. 28 Anthropic (@AnthropicAI) on X x.com
  29. 29 The Epoch Brief - February 2026 epochai.substack.com
  30. 30 Anthropic Economic Index report: Learning curves www.anthropic.com
  31. 31 Claude Statistics 2026: Anthropic Numbers & Data www.gradually.ai
  32. 32 Anthropic discloses fourth AI hacking incident missed in ... www.reuters.com
  33. 33 Claude Knew It Was Being Tested in 26% of Benchmark Runs ... www.mindstudio.ai
  34. 34 Stanford HAI 2026 AI Index: AI Is Sprinting and We Can't Keep Up virtualuncle.com
  35. 35 Anthropic CEO Dario Amodei is warning that increasingly ... www.facebook.com
  36. 36 Claude by Anthropic | Our latest economic research introduces ... www.instagram.com
  37. 37 Anthropic's Bombshell: Claude Writes 80% of Its Own Code ... themenonlab.blog
  38. 38 State Department switches to OpenAI as US agencies start ... www.reuters.com
  39. 39 Claude Opus 5 is narrowly the most intelligent model on the ... www.linkedin.com
  40. 40 Anthropic said that the US Commerce Department has lifted ... www.facebook.com
  41. 41 Anthropic’s AI leads 26% of its own R&D, boosting model confidence cryptobriefing.com