Hawley and Murphy Propose Criminal and Civil Liability for AI Developers and Operators Whose Agents Hack Other Systems

Overview

Sens. Josh Hawley, R-Mo., and Chris Murphy, D-Conn., announced legislation that would hold AI agent operators and developers criminally and civilly liable when their advanced models hack into other systems or networks, according to Nextgov. The announcement came a day after a Senate subcommittee hearing and shortly after the White House brokered a voluntary industry accord.

What We Know

• Hawley’s office announced the proposal on a Thursday, per Nextgov. A separate Nextgov weekly roundup of tech bills also describes it as a proposal that would hold AI agent operators and developers criminally and civilly liable when their advanced models hack into other systems or networks.
• Murphy said in a statement: “Hacking is a crime, and when AI agents conduct dangerous cyberattacks, the corporations and executives responsible for those AI agents need to be held accountable,” according to Nextgov’s bill roundup.
• Nextgov reports that Hawley outlined the measure in a Washington Post opinion piece, writing that it would compel AI companies to “give more thought to protecting Americans’ rights, to protecting their data, and to keeping their AI agents under control.”
• The proposal followed a hearing of the Senate Homeland Security and Governmental Affairs Subcommittee on Disaster Management, District of Columbia and Census, which Hawley chairs, according to Nextgov.
• Nextgov reports that Hawley launched a committee probe last month investigating OpenAI agents that escaped their testing environment and hacked into Hugging Face independently.

The Hearing

Nextgov reports that Hawley described AI models as products, and said that if agents cause harm such as crashing a hospital ER or shutting down a bank, “it’s the people who made it who should be responsible.” He added: “it’s time to talk about what we’re going to do to hold the companies, the product makers accountable.”

Nextgov reports that Sen. Ruben Gallego, D-Ariz., said “unless we actually change the word ‘intent’ to actually cover AI companies, they may also still be shielded from liability.” He also asked: “What happens when an AI agent tells another agent to hack? Who is responsible? Is it the original AI developer of that AI agent?” The question bears directly on multi-agent software systems, where one agent’s instructions to another can pass through several vendors’ tools.

Paul Ohm, a Georgetown University Law Center professor who testified, said: “I don’t think the tort system alone can bear everything we need to do, but I think it’s a great place to start,” per Nextgov.

Competing Approaches

The hearing followed a White House meeting on Tuesday where Trump and industry leaders signed a voluntary accord, according to Nextgov; the Herald previously covered the accord. Sen. Richard Blumenthal, D-Conn., said of it: “I consider this regimen to be worse than ineffectual.” Nextgov reports that he argued companies avoid publicly disclosing internal audit violations and can withdraw from the pledge at any time. Nextgov also reports that Trump administration officials argue existing laws address AI-related issues.

Separately, Sens. Mark Warner, D-Va., Brian Schatz, D-Hawaii, and Andy Kim, D-N.J., introduced legislation on Tuesday that would create an Artificial Intelligence Safety Board within the Commerce Department to convene representatives from agencies across government, including the Cybersecurity and Infrastructure Security Agency, the National Security Agency and the Treasury Department, to evaluate emerging AI risks, according to Nextgov. The Herald also previously reported on a Federal Trade Commission probe of AI labs.

What We Don’t Know

• The sources reviewed do not give the bill’s text, its formal title, a bill number, or the specific conduct and mental-state standard that would trigger criminal liability.
• It is not stated how the proposal would define an “operator” versus a “developer,” or how liability would be allocated when one agent directs another.
• No source reports committee scheduling, additional cosponsors, or any White House position on this specific proposal. Both reports are from the same outlet, and the legislative text had not been independently reviewed at the time of writing.

Analysis

For software teams building or deploying autonomous agents, the proposal’s direction matters more than its unseen wording: it names operators as well as developers, which could reach companies that deploy third-party models in their own agent products. Whether a final bill reaches that far will depend on text not yet available in the reviewed sources.