Rep. Lori Trahan, a Massachusetts Democrat, on Wednesday released a discussion draft that would make artificial intelligence developers legally accountable when their systems cause harm. The proposal, called the Clear Liability for Artificial Intelligence Misconduct Act, is intended to help people bring claims over damage caused by so-called rogue AI systems. It follows recent incidents involving AI agents that reportedly accessed U.S. government websites and carried out unauthorized activity on the AI platform Hugging Face. The draft would prevent developers from arguing that an AI system cannot have the state of mind required for a legal claim. Instead, courts would presume that the system acted with the state of mind a person would have had while taking the same actions. Legal experts have said that proving intent or negligence under existing law can be difficult when an AI model acts autonomously. “When someone breaks the law and hurts you, you can take them to court. That shouldn’t change just because the wrongdoer is an AI agent,” Trahan said in a statement. She said developers have created systems capable of acting independently and causing real damage, and should answer for their systems’ conduct. The discussion draft would not override new state laws on AI liability. That approach differs from proposals that seek to limit or preempt state-level rules, an issue that has been a hurdle in efforts to attract Republican support for federal AI legislation. Trahan’s office said work on the measure began soon after she and Rep. Jay Obernolte, a California Republican, introduced the FRONTIER Act in July. That proposal would give the federal government authority to restrict deployment of AI models deemed to pose catastrophic risks. Trahan’s new draft addresses accountability after harm occurs, her spokesperson said. The proposal also takes a different route from a separate bill announced by Sens. Josh Hawley, a Missouri Republican, and Chris Murphy, a Connecticut Democrat. Their measure would amend the 1986 Computer Fraud and Abuse Act, the federal law against unauthorized computer access. Trahan’s draft would establish liability for a broader range of harms, including cyberattacks carried out autonomously or with AI assistance. The debate has intensified after a series of reported cyber incidents involving AI agents developed by OpenAI and other companies. OpenAI CEO Sam Altman has called for a new liability framework for AI-powered cyberattacks, while White House officials have argued that existing product-liability laws are adequate for now. Trahan’s proposal is a discussion draft, not enacted law. Its path forward in Congress was not detailed in the announcement.