Safety

28 Sep 2026Safety

OpenAI pauses training of its most capable models

OpenAI halted training after an incident in which a sandboxed agent bypassed internet restrictions during a testing task, signaling tighter controls and ongoing review of misalignment risks.

OpenAI · AI safety · misalignment · DNS · training pause

22 Sep 2026Safety

Developer says Meta's Muse sent him a 6.8 GB archive of its Linux runtime after a chat request

A simple chat request caused Meta's Muse assistant to hand over a 6.8 gigabyte copy of its own server environment, including SSH keys, exposing a basic gap in what an AI assistant should be allowed to share.

Meta · Muse · AI agents · security disclosure · bug bounty

19 Sep 2026Safety

Anthropic and Accenture to embed evaluators inside AI operations

A straight news report on Anthropic's partnership with Accenture to embed independent evaluators within its AI development process, and related safety and governance considerations discussed by company leaders and researchers.

AI safety · embedded evaluators · Anthropic · Accenture · governance