The Trump administration, once committed to deregulating artificial intelligence, is now preparing to impose federal oversight on the development and deployment of advanced AI models. The White House is drafting an executive order that would establish a formal government role in vetting new AI technologies before they reach the market.
Shift in Policy
President Trump, who initially sought to free AI from government constraints, is now acknowledging the need for safeguards. This marks a stark reversal from his administration's earlier stance, which dismantled AI safety measures implemented by the previous Biden administration.
Sources confirm the White House is developing guidance to ensure federal agencies can safely adopt advanced AI models, such as Anthropic's Mythos and OpenAI's GPT-5.5.
Growing Concerns
The push for oversight follows concerns that AI has reached a threshold where its capabilities—particularly in cybersecurity—pose significant risks if left unchecked. The Pentagon recently labeled Anthropic as a 'supply chain risk,' prompting the White House to explore workarounds that allow federal use of its models.
In parallel, the National Security Council is crafting an AI security framework to mandate safety testing for AI models deployed by federal, state, and local governments. These measures aim to balance innovation with national security priorities.
Collaboration with Industry
Top AI labs, including Anthropic, Google, and OpenAI, are reportedly cooperating with the administration's efforts. While the Trump administration remains skeptical of regulation, it recognizes the necessity of partnering with industry to prevent more drastic measures.
The proposed executive order would create a working group of tech executives and U.S. officials to design oversight processes. Options under consideration include granting the government first access to new AI models without blocking their release.
Focus on Competition
The administration continues to prioritize beating China in the AI race, viewing it as an existential challenge. However, the breakneck pace of AI advancements has forced even the most committed deregulators to acknowledge the need for exceptions.