The White House has finalized a new framework designed to vet artificial intelligence models for cybersecurity and safety risks. This development follows months of discussions between administration officials and leaders from major technology firms, including OpenAI, Anthropic, Meta, Google, Nvidia, and Microsoft. Representatives from these companies attended a private meeting on Tuesday to review the finalized guidelines.

While the administration has moved forward with the framework, the specific details of the vetting process remain private. The lack of public disclosure has drawn criticism from observers who argue that the secrecy undermines transparency and may disproportionately benefit AI companies that prefer to keep their internal development processes shielded from public scrutiny.

There is a notable divide regarding the nature of the vetting process. While the administration has characterized the framework as a necessary step to mitigate potential dangers associated with advanced AI, the current implementation relies on a volunteer-based system. Critics suggest that without mandatory oversight or public transparency, the framework may fail to provide meaningful accountability for the industry. Conversely, proponents of the current approach argue that private collaboration is necessary to address complex technical risks without stifling innovation or compromising sensitive proprietary information. As of now, the administration has not provided a timeline for when or if the specific criteria used to evaluate these AI models will be made available for public review.