Trump signs AI order giving government access to powerful models

Lead

U.S. President Donald Trump signed an executive order on Tuesday creating a voluntary framework under which artificial intelligence developers will share advanced models with the government before public release. The order allows companies such as OpenAI, Google, and Anthropic to give the government access to their most powerful models for up to 30 days prior to planned release, according to reports from multiple outlets.

Coverage Comparison

Reuters and other news agencies reported the executive order's core provisions, including the 30-day preview window and the formation of an AI cybersecurity clearinghouse. Both reports highlighted that the order was triggered by concerns over Anthropic's Mythos model, which the company has withheld from public release due to its ability to expose vulnerabilities in computer systems, including those of banks, governments, and hospitals. The reports also noted that the 30-day window represents a compromise between an original draft that called for 90 days of pre-release access and tech companies' push for just 14 days.

While the two reports align on the main facts, one account includes a quote from Kent Walker, Google's head of public affairs, calling the order an "important step forward" that will ensure "defenders have the AI tools they need to keep America secure." The other report omits this quote but adds that David Sacks, Trump's AI and crypto czar, wrote on X that "unnecessary regulation is the biggest threat to innovation in America," and that winning the AI race required clearing "bureaucratic hurdles" from state legislatures and "woke" Washington politicians.

Key Claims

According to both reports, the executive order establishes a voluntary framework for AI developers to share advanced models with the government. The 30-day preview period is a compromise, as the original draft sought 90 days and tech companies pushed for 14 days. The order was motivated by Anthropic's Mythos model, which the company has not released publicly because it can expose vulnerabilities in critical systems. The order also directs the Treasury, the National Security Agency, and the CISA cybersecurity agency to form an "AI cybersecurity clearinghouse" in voluntary collaboration with industry and critical infrastructure operators.

Additionally, both reports note that Trump scrapped a Biden-era AI oversight order on his first day back in the White House. The Biden order, issued in 2023, required AI companies to share safety test results with the government and leaned heavily on voluntary commitments. In contrast, the EU's AI Act imposes binding rules for high-risk AI systems, including mandatory transparency requirements and safety testing obligations for the most powerful models.

Perspectives

Supporters of the executive order, such as Google's Kent Walker, view it as a positive step that strengthens national security by giving defenders access to AI tools. However, David Sacks, despite his role as AI and crypto czar, warned that the measure could slow innovation and hurt the U.S. in its AI race with China. Sacks' public comments suggest tension within the administration over the balance between security and innovation.

Critics, such as Anthony Aguirre, CEO of the Future of Life Institute, argue that voluntary frameworks are insufficient. Aguirre, quoted in one of the reports, said the government must be empowered to block the release of systems that pose an unacceptable national security risk. His remarks reflect a broader debate about whether voluntary measures can adequately address the dangers posed by advanced AI.

The order also raises questions about the divergence between U.S. and EU approaches. While the U.S. has opted for a voluntary framework, the EU's binding regulations impose stricter obligations on AI developers. This difference could have implications for global AI governance and for companies operating in multiple jurisdictions.

Background

The executive order comes after weeks of internal deliberation. According to Politico and other media, the White House appeared close to unveiling the measure earlier, only to pull back abruptly after Sacks called the president to warn that it would slow innovation. The final order represents a compromise, with the 30-day window shorter than the original 90-day proposal but longer than the 14 days sought by tech companies.

This order reverses the previous administration's approach. Trump scrapped a Biden-era executive order on AI oversight on his first day in office, signaling a shift toward less regulation. The new voluntary framework continues that trend, though the creation of an AI cybersecurity clearinghouse suggests some federal involvement in coordinating defenses against AI-related threats.

The EU's AI Act, which entered into force in 2024, takes a different approach. It imposes binding obligations on AI developers, including risk assessments and transparency requirements for high-risk systems. The contrast between the U.S. voluntary model and the EU's binding regulations will likely remain a point of contention as both regions seek to govern rapidly advancing AI technology.