Safety Evaluation Enters the Foundation-Model Stack
Google paired its robotics release with layered safeguards and an agentic-robot safety benchmark.
Semiconductor Domain: safety MCUs
Safe Agent Execution Gets Localized Obligations
HALO proposes localized obligations for admitting and supervising heterogeneous agent actions safely.
Trustworthy embodied intelligence gains a systems framework
A systems framework connects AI safety, control, cybersecurity, dependable computing, human factors, and lifecycle assurance.
