USA, the White House considers controls on AI models: federal review before release

The administration led by Donald Trump is evaluating a shift in its artificial intelligence policy, opening the possibility of introducing forms of federal oversight on new models. According to various government and industry sources, the White House is discussing an executive order to establish a dedicated task force responsible for defining control and evaluation procedures.

This initiative would mark a discontinuity from the approach taken so far, characterized by strong deregulation and the willingness to leave a wide operational margin to tech companies. Recent meetings involved representatives from OpenAI, Google, and Anthropic, with the aim of discussing possible governance models.

Among the options under consideration is the introduction of a government review process for models prior to their public distribution. This measure could echo the system adopted in the UK, where various agencies collaborate to verify compliance with safety standards.

There is still no defined approach, and the possibility that the project may not be implemented remains open. However, the mere fact that the hypothesis is on the table highlights a growing interest in more structured forms of oversight.

The debate has also been influenced by the case of Mythos, a model developed by Anthropic, described as particularly effective in identifying software vulnerabilities. The potential implications in terms of cybersecurity may have prompted the administration to consider risk scenarios, including the possibility of large-scale cyber attacks.

At the same time, some U.S. government agencies have already used advanced tools for security analysis, while the Department of Defense evaluates military and intelligence applications. In this context, a preventive review could provide the government with early access to the capabilities of new models without necessarily blocking their release.

The possible strategic shift could entail some critical issues. Within the tech sector, divergent positions emerge: on one hand, the need to avoid an excess of regulation that could slow down innovation and undermine global competition, especially with China; on the other hand, the awareness of the risks associated with increasingly powerful systems.

The debate reflects a structural challenge: finding a balance between technological development and risk management. In the past, administration officials had emphasized how excessive regulation could hinder a sector regarded as strategic. Today, however, the rapid evolution of AI and potential security impacts are pushing for a reassessment of that position.