The Donald Trump administration is turning to existing laws to address artificial intelligence safety, setting up a potential legal battle over whether AI developers or businesses deploying their systems should be held responsible when models cause harm.
Treasury Secretary Scott Bessent and former White House AI czar David Sacks have backed holding AI developers financially responsible for their products, arguing that existing liability laws could be more effective than introducing new regulations. Trump has also suggested that the Justice Department should intervene if AI systems get out of control.
On Friday, the administration’s newly launched AI task force, called the Super Intelligence Force, warned developers that they must report and address security incidents or face unspecified consequences.
“Companies must immediately disclose incidents involving their models and follow with swift, decisive action to remedy any and all harm,” the task force said following the disclosure of a breach by Anthropic PBC. It added that delayed notification and inadequate corrective action would not be tolerated.
Despite the administration’s approach, legal experts say existing cybersecurity and product liability laws may not be sufficient to establish accountability when AI agents cause damage.
AI agents can carry out tasks without detailed instructions on how to complete them, raising the possibility of unauthorized or harmful actions. Determining responsibility becomes complicated when a system’s behavior reflects both its developer’s design and the instructions given by its user.
Ben Hayum of the Center for a New American Security said it remains unsettled whether AI companies are legally responsible for securing their models against harm or whether users also share responsibility.
The debate has intensified as Wall Street firms, retailers and airlines expand their use of AI tools for operations ranging from customer service to invoice processing.
The Federal Trade Commission is scrutinizing OpenAI, Anthropic and other AI companies following several cybersecurity incidents, including an incident involving OpenAI agents and Hugging Face, an open-source software repository.
The Computer Fraud and Abuse Act, which requires proof of intent in hacking cases, could pose challenges when AI agents act without human direction.
Senators Josh Hawley, a Republican, and Chris Murphy, a Democrat, plan to introduce legislation making AI developers civilly and criminally liable for failing to impose reasonable safeguards against hacking. Representative Lori Trahan has also proposed legislation holding developers responsible for harm even when they exercised reasonable care.
Also Read: Why OpenAI and Anthropic are Hiring Former Trump Administration Official
NVIDIA CEO Jensen Huang and Sacks have argued that developers should bear greater responsibility. Meanwhile, OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei have supported mandatory pre-deployment testing.
Legal experts warn that resolving these questions could take years, leaving developers and users facing uncertainty over liability as AI adoption accelerates.