White House postpones executive order on AI
White House postpones executive order on AI – On Thursday, the White House announced the delay of President Donald Trump’s plan to sign an executive order aimed at creating a voluntary review process for artificial intelligence models before their release. According to a source with knowledge of the plans, the administration has pushed back the signing date, which was previously scheduled for Thursday afternoon. This marks the third time the order has been postponed, with Trump citing “certain aspects” of the document as a reason for the delay.
Reasons Behind the Delay
Trump expressed concerns that the executive order “gets in the way” of the United States’ leadership in technology and innovation. In a statement, he emphasized,
“we’re leading China, we’re leading everybody,”
suggesting that the proposed framework might act as a hindrance rather than a boost to the nation’s competitive edge. The president also noted that the order could have served as a “blocker,” reinforcing his stance that the delay is necessary to avoid potential obstacles to progress.
Voluntary Review Framework
The executive order, which had been under discussion for several months, was designed to require AI companies to voluntarily share advanced models with the government prior to public release. This provision, according to two sources, aimed to enable early scrutiny of models that could pose significant risks, such as those capable of enhancing cyberattack capabilities. The debate over the review period’s duration, however, remained a contentious point among stakeholders.
One version of the order proposed a 90-day pre-launch review period, while some industry representatives, including major AI firms, advocated for a shorter timeline, such as 14 days. The discrepancy in timeframes reflects differing priorities between the government and private sector: regulators sought more extensive oversight to identify potential threats, whereas companies argued for faster deployment to maintain market competitiveness.
Structure of the Executive Order
The order was divided into two key sections: cybersecurity and a framework for “covered frontier models.” The latter portion outlined criteria for determining which AI models would qualify for the voluntary review process. This classification was intended to streamline government evaluation of high-risk technologies, ensuring that only the most impactful models would undergo pre-public access scrutiny.
The cybersecurity section introduced a voluntary clearinghouse initiative, which would involve collaboration between the Treasury Department, other federal agencies, and AI firms. This body would work to detect and address vulnerabilities in unreleased models, providing an early warning system for potential security breaches. Additionally, the proposal called for expanded staffing at the U.S. Tech Force, a group tasked with modernizing government computer systems and enhancing their technological capabilities.
Industry Engagement and Early Access
Several leading AI companies, including OpenAI and Anthropic, had been actively engaging with the White House on the order’s development. Their collaboration extended beyond the executive order, with both firms granting select businesses and governments early access to their models. This approach, described by sources, was meant to fortify cyber defenses by allowing entities to test and adapt to new technologies before they become widely available.
Anthropic, for instance, unveiled its Mythos model earlier this month, which it claims can exploit cybersecurity vulnerabilities at an unprecedented rate. The company has opted to release the model through a restricted consortium of partners via its Project Glasswing initiative, rather than making it public immediately. OpenAI followed a similar strategy, offering early access to its latest AI models to help organizations prepare for potential threats.
Government’s Push for Oversight
Before the postponement, the Department of Commerce’s National Institute of Standards and Technology (NIST) had announced that major tech companies would share unreleased versions of their AI models with the government for evaluation. This move was framed as a step toward improving national security and public safety by identifying risks associated with emerging technologies. However, the announcement is no longer accessible on the Commerce Department’s website, indicating a shift in the administration’s public communication strategy.
Experts have long argued that advanced AI models can significantly amplify the scale and impact of cyber threats. By subjecting these models to early review, the government could take proactive measures to mitigate risks before they are deployed in real-world environments. Despite the delay, the administration remains committed to establishing a regulatory framework that balances innovation with security.
Broader Implications for AI Regulation
The Trump administration had previously adopted a hands-off approach to AI regulation, allowing companies to operate with minimal government intervention. However, recent developments, particularly the release of high-impact models like Mythos, have prompted a more cautious stance. The executive order represents an attempt to align federal oversight with the rapid pace of AI advancement, though its effectiveness has been called into question by some industry leaders.
While the voluntary nature of the review process was intended to encourage cooperation, critics argue that it may lack enforceability. Trump’s decision to delay the signing underscores the administration’s preference for flexibility in its regulatory approach, even as it acknowledges the importance of addressing cybersecurity concerns. The final version of the order, once signed, is expected to serve as a blueprint for how the government and private sector can work together to safeguard against AI-driven threats.
The White House has not yet provided a formal statement on the delay, leaving the public to speculate on the reasons behind the postponement. Nonetheless, the order’s components highlight a growing recognition of AI’s dual potential as both a tool for innovation and a vector for risk. As the debate continues, the administration’s approach may set a precedent for future AI regulation, shaping the landscape of technological development in the United States.
