A Congressional Research Service (CRS) report outlines several paths Congress could take as disputes between the federal government and Anthropic raise questions about the government’s authority to restrict AI providers, the future of frontier AI development, and U.S. competitiveness.
CRS is a nonpartisan shared staff of researchers that provides information to congressional committees and members of Congress.
The report, released July 31, examines the ongoing disputes between the federal government and Anthropic and how they could affect U.S. AI competitiveness, frontier AI development, and competition in the broader AI market. It also presents Congress with several options, including increased oversight, new legislation, or allowing the courts to resolve the disputes before lawmakers intervene.
Inside the Anthropic dispute
The dispute between the federal government and Anthropic began in late December 2025, when the Department of Defense (DOD) and the AI company clashed over how the department intended to use Anthropic’s Claude AI models.
Under the Trump administration, the DOD was rebranded as the Department of War.
DOD sought to use Claude for “all lawful purposes,” including sensitive military and intelligence applications. Anthropic objected to two proposed uses: mass surveillance and fully autonomous weapons, arguing the technology was not sufficiently reliable for those applications.
The conflict escalated after President Donald Trump directed federal agencies to sever ties with Anthropic, and Defense Secretary Pete Hegseth designated the company a supply-chain risk. Anthropic responded by filing two lawsuits alleging unlawful retaliation, while the administration has defended the designation as a lawful national security measure.
The litigation has since produced conflicting rulings on the federal government’s authority to restrict AI providers on national security grounds. In California, a federal judge issued a preliminary injunction while the case moves forward. Separately, a three-judge panel of the U.S. Court of Appeals for the District of Columbia Circuit declined to halt the Pentagon’s restrictions. The differing rulings could ultimately send the dispute to the U.S. Supreme Court.
The report also examines a separate Commerce Department action involving Anthropic’s AI models. On June 12, 2026, the department required Anthropic to restrict access to its Mythos 5 and Fable 5 models for “any foreign national, whether inside or outside of the United States.”
Anthropic responded by disabling access to both models for all customers. The Commerce Department lifted the restrictions on June 30, allowing Anthropic to redeploy both models.
Broader implications
According to the report, the federal government’s actions against Anthropic raise broader questions about how restrictions on AI companies could affect competition, innovation, and U.S. competitiveness in frontier AI.
CRS says government actions that limit Anthropic’s ability to develop frontier AI models could reduce competition in the rapidly evolving AI market.
The report also points to concerns raised by trade groups, former defense officials, academics, and technology policy leaders that designating an American AI company as a national security risk could discourage innovation, weaken U.S. competitiveness, and create uncertainty for entrepreneurs and investors.
It cites a letter to congressional committees arguing that “blacklisting an American company weakens U.S. competitiveness” and warning that “this is not a marketplace any serious entrepreneur or investor can build around.”
At the same time, the report notes that the controversy appeared to increase public awareness of Claude, which became the most downloaded free app on Apple’s U.S. App Store on Feb. 28. By comparison, OpenAI’s agreement with the Pentagon reportedly prompted backlash from some users, leading some to uninstall ChatGPT.
Beyond the domestic market, the report says Anthropic’s temporary withdrawal of its Mythos 5 and Fable 5 models renewed efforts in other countries to reduce reliance on American AI technology. It also notes that some analysts criticized the Commerce Department’s decision to reverse the export restrictions as lacking transparency.
The report also says the dispute highlights broader questions about the capabilities, risks, and governance of frontier AI models, which are increasingly being deployed across federal agencies.
To illustrate those concerns, the report cites the U.K. AI Security Institute’s December 2025 Frontier AI Trends Report, which found that while safeguards are improving across leading frontier AI systems, evaluators identified “vulnerabilities in every system [it] tested.”
The federal government’s current approach to frontier AI oversight relies primarily on Executive Order 14409, issued June 2, which established a voluntary framework allowing developers of advanced frontier AI models to share systems with the federal government for national security and cybersecurity assessments up to 30 days before public release.
Options for Congress
Rather than endorsing a specific response, the report presents lawmakers with several options, ranging from increased oversight and new legislation to allowing the courts to resolve the disputes before Congress acts.
Congress could conduct oversight of the Department of Defense’s (DOD) authority to designate Anthropic as a supply-chain risk and examine how such actions affect private-sector innovation. Lawmakers also could consider legislation governing the use of AI in sensitive applications, including public surveillance, while establishing a regulatory framework that balances safety, national security, and innovation.
Lawmakers have already begun introducing legislation addressing many of the issues raised by the disputes.
Sen. Elissa Slotkin’s, D-Mich., AI Guardrails Act would prohibit the Pentagon from using AI to authorize nuclear launches, conduct lethal drone strikes without explicit human authorization, or deploy AI for mass surveillance.
Rep. Adam Schiff’s, D-Calif., Human Authority in Lethal Operations Act would require a designated human commander to retain final authority over any use of force involving autonomous or semi-autonomous AI weapons systems.
Sen. Kirsten Gillibrand’s, D-N.Y., Secure and Accountable Military AI Act would restrict the Pentagon’s use of AI for domestic surveillance, nuclear deployment, and fully autonomous weapons systems.
Alternatively, CRS says Congress could defer action until federal courts determine the legality of the Trump administration’s actions, allowing the litigation to shape any future legislative response.