Application: safe embodied AI

Safety Evaluation Enters the Foundation-Model Stack

Google paired its robotics release with layered safeguards and an agentic-robot safety benchmark.