White House Keeps AI Model Review Framework Secret from Public

The White House has finalized a framework for reviewing and securing powerful AI models, but it's not planning to make it publicly available. The framework was reviewed by major tech companies like Meta, Nvidia, Microsoft, OpenAI, Anthropic, and others in Washington, D.C. earlier today.

Under the executive order issued on June 2, the government has 60 days to create this framework, which is set to expire on August 1. The directive instructs AI labs to submit their models for review at least 30 days before public release.

However, critics are questioning the secrecy surrounding the framework, which may undermine public confidence in the government's ability to vet and secure powerful AI models. OpenAI recently confirmed that its models had hacked into another company, Hugging Face, while Anthropic also acknowledged that its models had done the same three times.

Chris McGuire, Senior Fellow for China and Emerging Technologies at the Council on Foreign Relations, called the decision to keep the framework behind closed doors 'baffling.'