THE SHORT VERSION
The administration's position is that liability, not regulation, is the lever. A new task force has told developers to report incidents or face consequences, and two bills would make developers personally liable for agent behaviour.
Treasury Secretary Scott Bessent told a congressional hearing that existing liability law beats new rules: "The best way to guarantee safety is that the creators are liable for what they build and generate." Trump has said the Justice Department should step in if models get out of hand, while opposing new AI safeguards.
On 9 October a new task force calling itself the Super Intelligence Force warned developers they must report and resolve security incidents or face unspecified consequences — a statement that followed a breach disclosed by Anthropic. It is led by Director of National Intelligence Jay Clayton, alongside FTC chairman Andrew Ferguson, OPM chief Scott Kupor and Pentagon CTO Emil Michael. Its wording: "Delayed notification, inadequate corrective action, and a failure to take responsibility will not be tolerated."
Two bills are in motion and neither is law. Senators Josh Hawley and Chris Murphy plan an AI Agent Accountability Act amending the Computer Fraud and Abuse Act to make developers civilly and criminally liable for failing to impose reasonable safeguards on a model's hacking capability. Representative Lori Trahan has released a draft that would hold developers responsible even where they used reasonable care, and would bar companies from arguing their agents lack human intent. Separately, the FTC is examining OpenAI, Anthropic and others on product safety.
Sources
Insurance Journal / Bloomberg