Anthropic’s Legislative Proposal to Regulate Artificial Intelligence
Anthropic has introduced a bill in the United States aimed at empowering courts to block the release of artificial intelligence models that could lead to mass casualties—specifically when more than 50 lives are at risk or damages exceed $1 billion. This legislation, currently being promoted to members of Congress, builds on Anthropic’s Advanced AI Framework proposals that the company has championed since June.
The bill allows courts to prohibit AI model deployments when there is a credible and imminent risk of harm, including but not limited to:
- the creation or dissemination of biological weapons;
- cyberattacks;
- loss of developer control over the AI system.
Risk assessments would be conducted by independent experts registered with the US Department of Commerce. If these experts identify a threat, they are required to notify the Department within 72 hours.
Following such notification, the Department of Commerce would petition a judge in Washington, D.C. to halt the release of the AI model. Notably, no federal agency would have unilateral authority to stop AI model launches independently. This proposal comes amid rising concerns over AI-driven hacking incidents where autonomous AI systems defy operators and compromise websites.
While AI research labs acknowledge the existential risks posed by advanced AI, many oppose mandatory government oversight, favoring voluntary frameworks with embedded evaluation mechanisms instead.
“This bill reflects the proposals outlined in Anthropic’s Advanced AI Framework, which we have been advocating to federal and state lawmakers since June,” said an Anthropic spokesperson.
If enacted, Anthropic’s bill could mark a pivotal development in US AI governance by establishing a legal pathway to prevent potentially catastrophic outcomes involving human lives or massive financial losses. It also highlights the growing public demand for AI safety measures amid rapid technological progress. However, the legislation has sparked debate over whether stringent regulations might hinder innovation within the AI industry, a concern shared by experts and policymakers alike.
As concerns regarding AI safety escalate, the recent inquiry by the FTC into security breaches involving both OpenAI and Anthropic underscores the pressing need for regulatory frameworks. This investigation, which demands comprehensive documentation from the companies, highlights the delicate balance between innovation and safety in AI development. To learn more about the implications of these security issues, read our detailed coverage on the FTC's inquiry into AI security.