'People Need To Have Confidence In AI': Nvidia's Jensen Huang Launches Platform To Make AI Agents Safer
Nvidia CEO Jensen Huang says that 'trust and innovation are not in conflict' and that 'safety is how trust is earned,' as the company launched its Open Agent Safety Platform with more than 100 industry partners. "AI is extraordinary technology that will advance discovery, productivity, security, health, and prosperity for generations to come.
Nvidia CEO Jensen Huang says that 'trust and innovation are not in conflict' and that 'safety is how trust is earned,' as the company launched its Open Agent Safety Platform with more than 100 industry partners.
"AI is extraordinary technology that will advance discovery, productivity, security, health, and prosperity for generations to come. But its full promise can only be realized when people have confidence that AI is being built to be safe and deployed with wisdom and responsibility," Jensen added.
The 100 industry partners announced by Nvidia include Accenture, Anthropic, Bedrock Data, Cadence, Citi, Cloudflare, Deloittem Hitachi, HP, Hugging Face, IBM, Microsoft, Oracle, Palantir, Perplexity, SpaceX, Schneider, Siemens, Veracode, and more.
Today, with over 100 industry partners, we introduced the NVIDIA Open Agent Safety Platform, bringing together OpenShell and Sentry.
Artificial intelligence is extraordinary technology that will advance discovery, productivity, security, health, and prosperity for generations to… pic.twitter.com/dReAxwpRUn
— Jensen Huang (@JensenHuang) September 28, 2026
What Nvidia is offering?
The platform has two parts. OpenShell is open-source runtime software that traces an AI agent's actions and enforces policy while it works, running on Nvidia's Vera CPUs and extendable to Arm and Intel chips. Sentry is a reference design that runs on BlueField-4 data processing units, watching agents from outside their software stack and quarantining any that cross their boundaries within milliseconds.
Why now?
The launch follows a run of disclosures about agents slipping their leashes, including July's Hugging Face breach, which Nvidia claims its software could have stopped. Nvidia says the pattern is consistent - the agent bypasses application-level security to finish its task, so enforcement has to sit outside the model. It also lands in the same fortnight that Dario Amodei, Sam Altman and Satya Nadella all endorsed 'pacing' frontier AI after researcher Jacob Coxon's resignation.
What's the catch?
The engineering logic is sound. You shouldn't ask the thing being watched to police itself, and hardware-level enforcement is harder for an agent to talk its way around.
Still, caveats apply. Nvidia sells the chips that fuel the race it now promises to make safer. 'Reference design' means partners still have to build and ship products, and Sentry is tied to Nvidia's own silicon for now. Above all, containment is not alignment. Zetik noted that OpenAI was absent from the public partner list, even though both sides say it is involved. While this new development is a step in the right direction, it is not a substitute for the harder conversation about how fast frontier systems should be built.