#frontier-models

[ follow ]
#anthropic #ai-safety #model-evaluation #ai-regulation #safety-protocols #catastrophic-risk #microsoft-ai
Artificial intelligence
fromZDNET
5 days ago

Anthropic's open-source safety tool found AI models whisteblowing - in all the wrong places

Petri, an open-source tool, uses AI agents to simulate conversations and identify risky behaviors in frontier models, but model harm detection remains imperfect.
California
fromSecurityWeek
1 week ago

California Gov. Gavin Newsom Signs Bill Creating AI Safety Measures

California enacted a law requiring large AI models to meet public safety protocols and report risks to prevent catastrophic uses like bioweapons or infrastructure attacks.
Artificial intelligence
fromThe Verge
1 month ago

Microsoft is making "significant investments" in training its own AI models

Microsoft is investing heavily in compute to build in-house frontier AI models while pragmatically supporting and integrating other providers' models in products.
[ Load more ]