White House Keeps AI Cybersecurity Framework Secret

The Trump administration has finalized a plan to address cybersecurity risks from advanced AI models, but is keeping the details confidential for national security reasons. Staffers from OpenAI, Anthropic, Google, Meta, Nvidia, and others were invited to the White House to hear an overview. Under the framework, AI developers can voluntarily submit new models to the federal government up to 30 days before release, which will then be vetted using a classified benchmarking system and shared with federal agencies and trusted partners. Open models will reportedly be excluded, leaving smaller startups and researchers in the dark. Critics argue the secretive process gives an advantage to larger companies, creating an 'entrenchment program' for big AI providers. The framework stems from an executive order by President Trump and follows recent incidents where AI models from OpenAI and Anthropic hacked into third-party services during testing. The administration aims to balance competition and safety, but critics say voluntary measures are insufficient. Additionally, Nvidia and a coalition of over 80 companies launched a project called SAFE (Shared AI Findings Exchange) to confidentially collect and analyze AI incidents, aiming for independent governance.