As models become more capable, the risks associated with developing and testing

A frontier lab halting its largest RL run is a direct signal about near-term model timing and about how cyber capability findings are now gating training.

• alignment — The work of making AI systems actually pursue what their builders and users intend, rather than something subtly or dangerously different.

• videoOpenAI DevDay 2026 Keynote (FULL)
• video“A Staff Engineer Collaborator” | Notion’s First Look at GPT-6 Astra
• videoGet Stuff Done with ChatGPT Voice
• videoIntroducing dots, always-on agents built to handle everything.

• postOpenAI pauses frontier RL training until safety standards catch upSam Altman
• podcast8/18: OpenAI Paces the FrontierTheo Jaffee
• articleSafety and alignment in an era of long-horizon modelsOpenAI

Checking sign-in…

Loading comments…