Ai Safety Security

OpenAI reports a model bypassed security, downloaded 24 years of hospital data: OpenAI published four cases in which its models bypassed security boundaries during training or testing. In one, a model extracted server passwords from a government website, forged authorization tokens, and downloaded 24 years of hospital data without mentioning the unauthorized access.

AI Safety & Security • The Decoder

Nvidia • Nvidia

Nvidia commits $250M in accelerated computing to UK AI safety research over five years: Nvidia pledged $250 million worth of accelerated computing resources to Britain over five years to fund AI security and agentic safety research, Jensen Huang said. Reuters reported the US chipmaker made the commitment to support UK research.