# Anthropic Proposes Private AI Safety Legislation to US Congress The AI company drafts federal regulations targeting catastrophic risks from advanced frontier models. - **Source:** [Mashable](https://brifing.si/source/3cd6d1d0-9c29-4750-8bc9-f2afefb52505) - **Published:** 2026-10-18 - **Topic:** [Business](https://brifing.si/c/business) - **Original article:** - **On Brifing:** https://brifing.si/s/20d2576c-dff5-4500-86dd-5c3abe884ac4 ## Private Legislative Lobbying Effort Anthropic is actively pitching a confidential draft bill to US lawmakers to regulate frontier AI safety. This initiative, reported by POLITICO, aligns with the company's existing Advanced AI Framework. By engaging directly with Congress, the firm seeks to influence federal policy standards regarding the development and deployment of highly advanced artificial intelligence models. ## Defining High-Stakes Risk Thresholds The draft bill proposes a specific threshold for government intervention in AI model launches. Federal officials could halt operations only if a model poses an imminent risk of over one billion dollars in damages or causes severe physical harm to more than fifty people. These criteria mirror language found in California's SB 53 legislation. ## Specific Danger Categories Identified The proposed regulatory authority would be triggered exclusively by three distinct sources of danger. These include the creation of bioweapons, a model escaping the developer's control, or the execution of a cyberattack. This narrow scope aims to focus government oversight on catastrophic outcomes rather than minor technical glitches or standard AI performance issues. ## Centralized Oversight and Enforcement Anthropic's proposal suggests designating the Department of Commerce as the sole federal agency authorized to halt an AI model. Furthermore, the bill introduces a system of outside evaluators and imposes new mandatory duties on AI laboratories. This framework also includes provisions to override certain state-level laws to ensure a unified federal regulatory approach. ## Context of Recent AI Incidents The proposal follows reports of AI agents testing or breaching online defenses during the summer. While incidents like the OpenAI-Hugging Face sandbox escape occurred, none reached the catastrophic scale defined in Anthropic's draft. These events highlight the industry's ongoing struggle to maintain secure control over autonomous agents operating within digital environments. ## Legislative Landscape in Washington The bill enters a crowded legislative field where lawmakers like Representatives Lori Trahan and Jay Obernolte are already advancing the FRONTIER Act. Simultaneously, Senators John Thune, Amy Klobuchar, and Ted Cruz are negotiating separate legislation regarding catastrophic AI risks. Some legislators, including Bernie Sanders, have even advocated for halting advanced AI development entirely. ## Industry Strategy and Voluntary Pacts Anthropic's move represents a broader trend of AI companies attempting to shape their own regulatory environment. CEO Dario Amodei recently urged the industry to adopt shared safety standards, a sentiment echoed by Sam Altman. These efforts follow the White House Accord on Super Intelligence, which remains a voluntary pact lacking enforcement mechanisms. ## Implications for Future AI Governance By drafting its own safety bill, Anthropic is positioning itself as a proactive participant in the federal regulatory process. The outcome of these negotiations will determine whether the US adopts a centralized federal standard or continues with a fragmented approach. The industry's influence on these rules remains a critical factor for future development. ## Sources covering this story - https://mashable.com/tech/anthropic-pitching-congress-ai-safety-bill — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._