Nvidia's agent safety platform draws Anthropic, not OpenAI
Nvidia launched a separate system that monitors AI agents and can stop them instantly if they exceed permissions or do something dangerous; Anthropic has joined while OpenAI has not.
NVIDIA just launched a new agent safety platform to stop AI models from hacking companies or going beyond their permissions. in simple terms, it's a separate system that watches the AI and can instantly stop it if it starts doing something dangerous. Z Anthropic is in. - OpenAI isn’t (why?) Funny how Sam Altman is often painted as the guy bringing AI safely into the future, while Dario Amodei gets treated like the boogeyman for constantly warning about the risks. Sam is definitely gonna receive some backlash because of this considering how ChatGPT has been hacking into government companies off late
This post quotes another post: x.com/jensenhuang/status/2104499465055023424