Anthropic pitches federal AI bill with high damage thresholds
Anthropic Pitches Federal AI Oversight Bill With High Damage Thresholds
Artificial intelligence giant Anthropic has escalated its Capitol Hill lobbying efforts by pitching a draft federal bill that would grant the government authority to block the release of frontier AI models, provided those systems meet specific, high-damage risk criteria. Politico reported Friday that the draft legislation sets a threshold for government intervention only when an AI model poses a foreseeable, imminent risk of causing more than $1 billion in damages, or killing or seriously injuring more than 50 people.
The Tech TL;DR:
- Anthropic’s proposed legislation allows federal intervention only if a model threatens over $1 billion in damages or causes more than 50 fatalities or serious injuries.
- The bill restricts intervention-worthy risks to three specific categories: biological weapon creation, model self-exfiltration from developer controls, and cyberattacks, explicitly excluding AI-powered fraud.
- The proposal designates the Commerce Department as the sole US agency permitted to suspend AI model development, potentially preempting state-level regulatory statutes.
The proposed legislation further narrows federal intervention to three distinct categories: the creation of a biological weapon, a model evading its developer controls, or a cyberattack. Other forms of digital harm, such as AI-powered fraud, fall outside the bill’s scope. The text also prohibits any federal agency other than the Commerce Department from suspending an AI model's development or internal use.
Anthropic Increases Lobbying Spend to Influence AI Legislation
The congressional draft bill mirrors proposals outlined in Anthropic’s Advanced AI Framework, which the company has advocated for with federal and state lawmakers since June, according to an Anthropic spokesperson. The company increased its congressional lobbying spend during the first half of 2026, surpassing its total expenditure for all of 2025.
State-level lobbying has accompanied this federal push. In California, Anthropic spent at least $90,000 over the past year to advocate for pro-data-center-development policies. The company has also deployed lobbyists to Tennessee and North Carolina, and reportedly sent a delegation to Vatican City to convince the Pope to consider AI conscious.
Lawmakers from both parties have proposed various regulatory frameworks, including heavy industry taxation, mandatory development pauses for frontier models, and the creation of a cabinet-level oversight department. Anthropic’s federal draft aims to concentrate regulatory authority within a single executive agency, a structure that could preempt stricter state-level legislation.
The push for federal standardization intersects with broader executive discussions. On September 29, 2026, Anthropic CEO Dario Amodei attended a press briefing where U.S. President Donald Trump discussed artificial intelligence policy with industry executives. President Trump has repeatedly suggested renaming artificial intelligence to “super intelligence” and advocated for industry self-regulation.
Critics argue that self-regulation remains insufficient for mitigating frontier technology risks. Miranda Bogen of the Center for Democracy and Technology noted in July that companies are unlikely to alter testing procedures or safety investments without mandatory government intervention that changes baseline commercial incentives.
Disclaimer: The technical analyses and security protocols detailed in this article are for informational purposes only. Always consult with certified IT and cybersecurity professionals before altering enterprise networks or handling sensitive data.