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

Nvidia announced a new consortium of over 100 companies dedicated to addressing rogue AI agents, but OpenAI notably did not publicly join despite its rival Anthropic becoming a supporter. OpenAI is working with Nvidia on agent security components including OpenShell, open source software designed to prevent agents from escaping sandboxes, and a spokesperson confirmed the company supports Nvidia's work even without a public pledge. The Open Agent Safety Platform also uses proprietary hardware monitoring through Nvidia's BlueField-4 processors that can observe and shut down agents without their detection, which may discourage competitors and companies seeking independence from Nvidia from joining publicly. OpenAI appears to be pursuing its own separate approach to AI safety through internal safeguards, its own cybersecurity consortium called the Defense Factory, and commercial security offerings, positioning AI safety as a business opportunity rather than a shared industry standard.