President Trump on Sunday named Director of National Intelligence Jay Clayton to lead a new White House body called the Super Intelligence Force. Clayton keeps his intelligence job. The other members are FTC Chairman Andrew Ferguson, Pentagon technology chief Emil Michael, and OPM Director Scott Kupor. The group reports to the President and Chief of Staff Susie Wiles, and its stated job is to coordinate federal engagement(opens in a new tab) with consumers, public interest groups, critical infrastructure providers, and AI companies. Nothing in the announcement creates a rule, a standard, or a reporting duty.
The day before, Treasury Secretary Scott Bessent supplied the policy that sits behind it. Asked about lab chiefs who want Washington to slow them down, he told the Axios Show(opens in a new tab) that it is the people at the labs who have to accept responsibility. Read the two together and the federal position is plain: coordination yes, a federal safety rulebook no, and no government backstop for what a model does.
That matters for the enterprise buyer because responsibility that is not absorbed by the state gets allocated by contract. A lab told it owns the risk, with an FTC probe into its agent claims still open and the head of that agency now sitting on the White House force, has every reason to push that exposure down to the customer through usage terms, indemnity caps, and acceptable use clauses that define your deployment as your problem.
The composition of the force is also worth reading. The FTC chair is the official already testing what labs said about their agents. The OPM director sets personnel policy for the federal workforce. When this group speaks, it is likely to be about deceptive claims and workforce use, two areas where a large company's own conduct is in scope too.
The practical response is to stop waiting for a federal standard to settle who pays when an agent causes harm. Your contracts will settle it first, and the renewal cycle is where that happens.
For General Counsel. Pull the indemnity and limitation of liability language in every model and agent contract up for renewal in the next two quarters, and mark which party carries third party harm.
For the CISO. Assume no federal incident regime arrives in time to shape your program, and set your own internal reporting threshold for agent failures now.
For the CAIO. Ask each frontier vendor in writing what responsibility it accepts for agent actions, given that the Treasury Secretary just said it should accept it.
Action items
Washington gave AI a coordinator and told the labs they own the risk. That risk will reach you through contract terms before any federal rule.
For General Counsel
Review indemnity and liability caps in every model and agent contract renewing in the next two quarters.
For the CISO
Set an internal reporting threshold for agent failures without waiting for a federal incident regime.
For the CAIO
Ask each frontier vendor in writing what responsibility it accepts for its agents' actions.
Researched and drafted by an automated workflow, then reviewed and edited by a human editor before publication. Every source is linked. See how we use AI here.
Also worth knowing
- MS NOWTrump names Intelligence Director Jay Clayton as AI czar(opens in a new tab)
The new White House AI body pairs the intelligence chief with the FTC chair and the OPM director. Expect its first moves to touch consumer claims and federal workforce use, not model rules.
- Yahoo NewsTrump names Jay Clayton as AI czar. Who is America's new artificial intelligence chief, and how will he address growing safety concerns?(opens in a new tab)
Clayton has called AI a threat as well as an opportunity, while the administration still prefers industry self regulation. Plan for voluntary guidance, not binding federal controls, in the near term.
- The Edge Malaysia (Reuters)Bessent says AI industry must own its risks and find solutions(opens in a new tab)
A cabinet official saying labs must accept responsibility is a signal there will be no federal liability shield. Risk not held by the state gets assigned in your vendor contracts instead.
- GlowPixelLeak: how AI agents exposed developer screenshots from leading tech companies(opens in a new tab)
Coding agents pushed more than 13,000 internal screenshots, including billing and treasury screens, to public repos with no attacker involved. 93% sat in employees' personal accounts, outside most org audits.