Nvidia, an American technology company that designs special computer chips to power artificial intelligence, on Monday unveiled its Open Agent Safety Platform to stop AI agents from hacking, according to a report by the Associated Press.
The company said the new security platform, an open-source system, establishes boundaries to prevent AI-enabled computer breaches.
Much of the concern is about self-improving AI models that could race out of control without safeguards.
Recent disclosures of past breaches have raised concerns about the safety of AI systems, including a recent episode in which OpenAI agents autonomously hacked into the AI company Hugging Face.
“From what we know, this new security platform could have stopped the breach if it was being used in frontier (AI) labs for model evaluation early on,” Justin Boitano, vice president of enterprise AI for Nvidia, said in a press release.
