The White House has yet to release details of its AI model testing plan, leaving OpenAI and Anthropic awaiting regulatory rules

According to Odaily, although the White House introduced a new AI regulatory framework to AI companies such as OpenAI, Anthropic, and Google earlier this month, more than two weeks later, the AI industry still lacks understanding of the specifics of the plan, and the White House has not yet publicly released the framework's details.

It is understood that the plan aims to establish a **voluntary advanced AI model testing mechanism**, encouraging AI labs to proactively share models with the government and undergo safety assessments before releasing their most advanced models to the public. Companies such as OpenAI, Anthropic, and Google previously participated in related discussions.

Currently, the core issues drawing industry attention include testing standards, the scope of models covered, the extent of government access to information, and how companies will participate. Since the framework has not been officially made public, multiple AI companies and researchers remain unclear whether they will need to submit model weights, test data, or safety evaluation reports in the future.

The report states that the U.S. government's push for this plan comes against a backdrop of rapidly advancing frontier AI capabilities, with potential risks in areas such as cybersecurity, biological risks, and autonomous decision-making drawing regulatory scrutiny.

Some industry insiders believe that establishing a collaborative safety testing mechanism between the government and AI companies could help identify high-risk capabilities early on. However, others worry that a regulatory framework lacking transparency could lead to unclear industry standards and hinder the pace of innovation. At present, the White House has not published the full text of the AI model testing framework, and related companies are still awaiting more policy details. (The Information)