OpenAI skips Nvidia's rogue AI safety consortium

Nvidia launched the Open Agent Safety Platform with over 100 company supporters to combat rogue AI agents, but OpenAI notably did not join publicly, despite rivals Anthropic signing on. An OpenAI spokesperson said the company is supportive and is collaborating on OpenShell, a key sandbox component. OpenAI's absence may reflect a desire for independence from Nvidia, its major investor, and a push to lead its own safety efforts. The platform includes proprietary Nvidia hardware called Sentry, which may deter competitors. OpenAI is separately building its own cybersecurity products and runs its own information-sharing consortium, the Defense Factory.