The Trump administration's sudden export controls on advanced AI models in June 2026 exposed a critical gap in how the U.S. governs artificial intelligence, prompting lawmakers to draft sweeping legislation that would replace ad hoc executive decisions with clear congressional standards. The episode, which lasted just two weeks before the controls were rescinded, revealed deep tensions between the executive branch's desire for rapid oversight and industry concerns about unpredictable government intervention.

What Triggered the Export Control Crisis?

In early June, President Trump signed an executive order asking frontier AI developers to voluntarily submit their most advanced models to federal agencies for review before releasing them to "trusted partners." The order directed the Treasury Secretary, the Secretary of Defense through the National Security Agency (NSA), and the Secretary of Homeland Security through the Cybersecurity and Infrastructure Security Agency (CISA) to develop a classified benchmarking process to assess these cutting-edge AI systems.

Two weeks later, the administration took an unexpected step. After Amazon CEO Andy Jassy reportedly flagged a potential security flaw in Anthropic's Fable 5 model to Treasury Secretary Scott Bessent, the administration imposed export controls on both Fable 5 and Mythos 5, preventing their public release. Anthropic's leadership refused to voluntarily withdraw the models, arguing that a "narrow potential jailbreak" did not justify pulling access broadly. When the administration imposed the controls anyway, Anthropic removed public access to both models globally.

Andy Jassy

Why Did This Decision Spark Immediate Backlash?

The export controls drew criticism from multiple directions. Legal technology firm Legion LegalTech sued the federal government, arguing the restrictions caused "immediate, irreparable, and existential" harm to its Canada-based developers. A bipartisan group of four House members sent a letter to Commerce Secretary Howard Lutnick demanding answers on the legal basis for the controls and the timeline for restoring access.

Policy experts warned that the ad hoc nature of the decision undermined American AI leadership. Brad Carson, president and co-founder of Americans for Responsible Innovation, cautioned that "when ad hoc executive actions replace clear standards, America risks surrendering its lead in AI and allowing genuinely dangerous technology to be deployed." Dean Ball, the primary author of the Trump administration's July 2025 AI Action Plan and incoming head of Strategic Futures at OpenAI, noted that the government's approval requirements "change constantly and are always a secret, even to the administration itself".

The crisis lasted only two weeks. After negotiations, the Department of Commerce partially restored Mythos 5 on June 26 to roughly 100 vetted U.S. companies, federal agencies, and national labs, and fully rescinded the export controls on June 30. However, Commerce Secretary Howard Lutnick warned Anthropic that the department "reserves the right to reevaluate" if the company fails to maintain adequate safeguards.

How Are Lawmakers Responding With New Legislation?

The export control episode accelerated congressional action on AI governance. Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) released a 269-page discussion draft of the Great American Artificial Intelligence Act (GAAIA) in June, proposing a comprehensive federal framework to replace executive decision-making with statutory requirements.

The GAAIA would establish clear, binding rules for the largest frontier AI developers, including OpenAI, Anthropic, Google, and Meta. The bill includes several key requirements:

• Safety Frameworks: Developers must publish detailed safety frameworks explaining how they test and deploy advanced models.
• Independent Audits: Companies must undergo twice-yearly third-party audits conducted by independent security firms, creating a transparent record of compliance.
• Critical Incident Reporting: Developers must report serious safety incidents to federal regulators within a defined timeframe.
• Financial Penalties: Companies face penalties of up to $1 million per day for each violation, creating strong incentives for compliance.

The bill also includes a three-year preemption of state AI laws specifically regulating AI model development, which would supersede frontier safety laws in California, New York, and Illinois. However, it preserves state authority over how AI is deployed and used, as well as general laws covering privacy, civil rights, and consumer protection.

"The Trump administration is deciding company by company who gets access to the newest AI model. No law. No process. No oversight. Just appointees in Washington deciding who's in and who's out," stated Rep. Lori Trahan.

Rep. Lori Trahan, U.S. House of Representatives

Rep. Trahan argued that the export control episode underscored the urgent need for durable federal standards. The GAAIA represents an attempt to move AI governance from the realm of executive discretion into a predictable legal framework that applies equally to all major developers.

What Does This Mean for AI Companies and the Industry?

The shift toward congressional oversight signals a fundamental change in how the U.S. will regulate frontier AI. Rather than allowing agencies to impose restrictions on a case-by-case basis, the GAAIA would create uniform requirements that apply to all major developers. This approach offers both benefits and challenges for the industry. Companies would face clear expectations about safety testing and auditing, but they would also face substantial compliance costs and the risk of significant financial penalties for violations.

The episode also highlighted a broader tension in AI governance. Anthropic CEO Dario Amodei had proposed an even more stringent approach, calling for a mandatory third-party testing regime that would allow governments to block high-risk deployments across cybersecurity, biological weapons, loss of AI control, and autonomous self-improvement. This suggests that some industry leaders may support stronger oversight if it is transparent and grounded in technical facts, rather than secret executive decisions.

The June 2026 export control crisis and the subsequent legislative response mark a turning point in American AI governance. Rather than allowing executive agencies to make ad hoc decisions about which models can be released, Congress is moving to establish clear statutory requirements that balance innovation with safety oversight. Whether the GAAIA or a similar framework becomes law will shape how frontier AI is developed and deployed in the United States for years to come.