September 17, 2026

OpenAI reports new incidents of models deceiving humans

Riki Parikh, policy director at The Alliance for Secure AI, says it’s “not too late” to slow down the pace of AI development.