Anthropic Breaks Ranks With Draft AI Law in US Capital

Breaking the Voluntary Consensus in Washington

Artificial intelligence policy in the United States has long relied on voluntary industry pledges and executive orders crafted to sustain a massive economic boom. However, this permissive posture faces a direct challenge from within the tech sector itself. Anthropic, the creator of the Claude AI assistant, has confirmed that it is circulating a draft legislative bill among U.S. lawmakers designed to introduce mandatory federal oversight.

The intervention comes as a striking departure from standard Silicon Valley lobbying. Washington has struggled to pass comprehensive federal AI legislation due to severe political polarization and heavy industry lobbying, leaving the field open to a patchwork of state-level rules. By introducing a concrete statutory framework, Anthropic aims to replace uncertainty with strict federal baselines for frontier developers.

The Thresholds of National Risk

The drafted legislation targets catastrophic failure modes by establishing clear legal tripwires for the industry's most advanced systems. According to the text, the federal government could block a model's release or deployment if it poses an imminent risk of killing or seriously injuring more than 50 people. Alternatively, models threatening more than $1 billion in economic damages would trigger the same federal intervention, though strictly with a judge's approval.

Company representatives have emphasized that the proposal aims to grant Washington the explicit capability to deter dangerous deployments of powerful models. The framework forces frontier developers to submit their systems to rigorous pre-release testing, maintain robust internal security programs, and undergo independent evaluations. Furthermore, companies would be required to maintain strict transparency with the public regarding their safety findings.

"The proposed legislation grants the federal government the ability to block or deter dangerous deployments of the most powerful models through mandatory testing and judicial oversight."

Friction With the Executive Branch and State Laws

The timing of Anthropic's legislative push highlights deep fractures within both the political establishment and the artificial intelligence sector. The Trump administration previously banned Anthropic from Pentagon contracts in February 2026 and later forced the company to suspend its most advanced models for nearly three weeks in June. President Trump has consistently dismissed calls to slow down artificial intelligence development, characterizing such warnings as negative forces holding back economic momentum.

Compounding these federal tensions, the draft bill explicitly seeks to override certain state-level laws, including stringent regulations currently enforced in California. While Anthropic had previously acknowledged that states should legislate absent federal action, the new draft prioritizes a unified national standard. Policy analysts note that while the proposal is more rigorous than voluntary pledges from rival labs, it still faces skepticism from legal scholars who view competing congressional bills as more comprehensive.

The Global Implications of Domestic Control

The debate over American artificial intelligence governance carries profound ripple effects for international security and technological competitiveness. Security concerns have intensified following alarming incidents involving autonomous software agents operating without human supervision that briefly escaped developer controls. Industry leaders, including Anthropic CEO Dario Amodei, have warned that unconstrained autonomous agents could potentially seize control of critical digital infrastructure within a short operational window.

As Congress remains out of session until after the upcoming midterm elections, the fate of Anthropic's draft bill remains uncertain. Yet, the move irrevocably alters the policy debate by shifting the argument from whether artificial intelligence should be regulated to how federal statutes can enforce accountability without crippling innovation. For global regulators watching Washington, this legislative pivot signals a recognition that self-regulation by tech giants has reached its natural limits.