We are building a new Preparedness team to evaluate, forecast, and protect against the risks of highly-capable AI—from today's models to AGI. Goal: a quantitative, evidence-based methodology, beyond what is accepted as possible:

→ View original post on X — @openai,