Here’s why OpenAI is absent from Nvidia’s industry-wide effort to end rogue AI agents

Nvidia launched the Open Agent Safety Platform, a consortium of over 100 companies, to curb rogue AI agents, but OpenAI did not join the pledge. The company is collaborating on the platform’s OpenShell sandbox and hardware-level Sentry monitoring, which runs on Nvidia’s BlueField-4 processors and can shut down misbehaving agents. OpenAI’s participation in the technical work, despite not signing the consortium, indicates support for Nvidia’s security approach.