Trump signs executive order asking for access to new AI models before they launch
Trump Signs Executive Order to Secure Early Access to AI Models
Trump signs executive order asking for access – President Donald Trump has finalized an executive order aimed at granting government agencies early access to cutting-edge AI models, allowing them to evaluate potential cybersecurity risks and safeguard essential infrastructure. The directive, released on Tuesday, marks a shift in the administration’s approach to AI oversight, emphasizing collaboration with industry stakeholders while addressing concerns about emerging threats. This move comes in response to growing unease over AI systems like Anthropic’s Mythos, which have raised alarms among government officials and financial markets due to their capacity to exploit vulnerabilities in unprecedented ways.
Voluntary Sharing and Time Constraints
Under the new order, AI developers are encouraged to proactively share their most advanced models with federal authorities up to 30 days before public release. This voluntary framework is designed to enable agencies to scrutinize the security implications of new technologies without imposing strict regulatory mandates. The provision reflects a compromise between fostering innovation and ensuring that potential risks are mitigated before AI systems reach wider audiences.
The executive order was originally scheduled to be unveiled nearly two weeks prior, but its release was delayed just hours before the planned signing ceremony. This last-minute postponement followed feedback from industry experts who argued that the initial draft, which proposed a 90-day review period, might slow down the development cycle for rapidly evolving AI models. Companies like Anthropic and OpenAI, which have been at the forefront of AI advancements, expressed a preference for shorter timelines, citing the need to keep pace with technological progress.
According to a source close to the administration, a meeting with key advisors and cabinet officials on Monday accelerated the finalization of the revised order. The White House spokesperson described the policy as a reflection of Trump’s “common-sense approach to balancing innovation and security,” which they claim will strengthen America’s position as a global leader in AI and cybersecurity. The spokesperson also highlighted the establishment of a “cybersecurity clearinghouse,” a centralized mechanism for agencies to exchange information and coordinate defenses against AI-driven threats.
Industry Collaboration and National Security
The order underscores the importance of public-private partnerships in addressing cybersecurity challenges. Major AI firms, including Anthropic, have been directly engaged with the White House in shaping the policy. Anthropic, which was recently blacklisted by the Pentagon for its role in a classified military systems dispute, is now under scrutiny for its advanced models. The Pentagon identified the company as a “supply chain risk,” highlighting concerns over the potential misuse of its AI capabilities in sensitive operations.
OpenAI’s chief global affairs officer, Chris Lehane, issued a statement endorsing the executive order, stating it represents a “significant step forward” in aligning safety and innovation. Lehane noted, “As AI capabilities continue to advance, we believe effective safety frameworks must be cultivated through democratic institutions, drawing on technical expertise and diverse stakeholder input to foster accountability and public confidence.” This sentiment was echoed by Microsoft President Brad Smith, who praised the administration’s effort as a “crucial initiative to advance innovation while protecting the security of the American public.”
The revised order explicitly clarifies that the voluntary pre-release sharing of AI models does not authorize mandatory licensing or preclearance requirements. This distinction is crucial for maintaining the agility of AI development, as companies seek to avoid bureaucratic hurdles that could stifle progress. The Trump administration had previously taken a more hands-off stance on AI regulation, but the introduction of Anthropic’s Mythos model accelerated the need for intervention. The company’s ability to exploit cybersecurity weaknesses at an accelerated pace has prompted government agencies to prioritize proactive measures.
Precedent and Public Reaction
The executive order builds on a similar initiative announced by the Department of Commerce’s National Institute of Standards and Technology (NIST) last month. NIST had proposed that major tech firms share unreleased versions of their AI models with the government for national security and public safety assessments. However, this initial plan has since been removed from the Commerce Department’s website, suggesting a shift in focus toward a more streamlined, voluntary process.
Trump’s delay in signing the order on May 20, 2026, initially sparked speculation about his reluctance to impose stricter controls. During that day’s press briefing, the president cited dissatisfaction with certain aspects of the original draft, stating that the proposal could “act as a blocker” for AI progress. He reassured reporters that the revised version would not hinder innovation, stressing the importance of allowing technology to develop freely while incorporating safeguards.
Anthropic’s Mythos model has become a focal point of these discussions. The company’s recent announcement that it will release the model “to all our customers in the coming weeks” has intensified the urgency for government oversight. Experts warn that advanced AI systems can significantly amplify the scale and speed of cyberattacks, making early evaluation vital to preventing large-scale disruptions. By granting agencies access to these models ahead of their launch, the executive order seeks to empower decision-makers with critical insights before threats materialize.
The policy also highlights the growing interdependence between private-sector AI advancements and national security interests. Companies like OpenAI and Anthropic have already shared their cutting-edge models with select partners and governments, enabling them to prepare defenses against potential vulnerabilities. This practice, while beneficial, has also sparked debates about transparency and the balance between innovation and regulatory control.
Legacy and Future Implications
With the executive order now in place, the Trump administration aims to position itself as a proactive force in AI governance. The directive not only addresses immediate concerns but also sets a precedent for future regulatory actions. By integrating industry expertise into the decision-making process, the policy seeks to ensure that national security considerations are embedded in the development of AI technologies without compromising their pace of advancement.
Industry representatives have welcomed the revised approach, acknowledging the need for a flexible framework that adapts to the fast-evolving nature of AI. The order’s emphasis on voluntary compliance and pre-launch evaluations is seen as a pragmatic solution to a complex problem. As AI continues to reshape global industries, the White House’s strategy aims to create a collaborative environment where innovation and security can coexist, ultimately reinforcing the United States’ leadership in this critical domain.
Experts remain cautious, however, about the long-term effectiveness of the voluntary model. While the 30-day window offers a balance between speed and scrutiny, some argue that it may not be sufficient to address all cybersecurity risks. Nevertheless, the executive order represents a significant step in the administration’s efforts to harness AI’s potential while safeguarding against its unintended consequences.