All top stories · This edition

Nvidia unveils AI security platform to block rogue agents

September 28, 2026

Summary

Nvidia introduced new open-source software security tools to prevent AI agents from misbehaving, claiming they could have stopped the Hugging Face hack. Meanwhile, OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei will not attend an Australian Senate inquiry into AI, though Trump confirmed a dinner meeting with Anthropic's CEO.

Coverage by lane

Left

  • Nvidia announces security system to stop AI agents from going rogue
  • Nvidia unveils security platform to stop AI agents from going rogue

Center

  • Nvidia releases software platform to stop AI agents from misbehaving
  • Nvidia Rolls Out New Tools to Keep AI Agents in Line
  • OpenAI, Anthropic CEOs Will Not Attend Australia AI Inquiry

More from this edition

  • UK arrests five near US-used air base on terror suspicion; Iran link probed
  • Pope Leo XIV praises European democracy, decries sexual abuse and assisted dying
  • Hurricane Polo threatens Mexico's Baja California; Nolo skirts Hawaii
  • South Korea Demands Ukraine Apology Over North Korean POW Disclosure
  • Channel crossing attempt kills three, including child, off French coast
  • US, China Agree to Cut $60 Billion in Tariffs
  • India mosque demolitions spark police clashes; rape accountability urged
  • Trump rejects Iran's Strait of Hormuz proposal; oil prices rise