The White House's artificial intelligence task force agreed at its first meeting to build a way for AI developers to warn the government about breaches, hacks and jailbreaks, and to commission a risk report from the president's science advisers, Bloomberg reported Friday. The plan is the first sign of how the administration intends to turn its demand for AI incident disclosure into a working process, though members discussed no new laws or rules to back it.

What Happened

The group, which the administration calls the Super Intelligence Force and which Director of National Intelligence Jay Clayton leads, met for the first time Thursday. Members agreed to focus on AI risks in three areas: cybersecurity, biosecurity and what the group calls model control and accountability, a category that broadly concerns whether developers can keep powerful systems behaving as intended and answer for them when they do not.

According to a White House account of the meeting cited by Bloomberg, the task force will ask the President's Council of Advisors on Science and Technology, a panel of outside experts, for a report assessing those and other concerns. Bloomberg's report did not give a deadline for that work.

RelatedUK Weighs AI Incident Reporting Rules and Tells Labs to Hold Back Systems They Cannot Control

The same account says the task force will develop protocols and mechanisms for labs to provide "early warnings and systemized notifications" to the government about breaches, hacks, jailbreaks and other problems. A jailbreak is a technique that tricks an AI model into ignoring its safety limits.

The group will also set up a law enforcement and consumer protection coordination committee that includes the Justice Department and the Federal Trade Commission, among other agencies. It plans to consult frontier model developers, which build the most capable AI systems, along with other AI companies, consumer protection and civil liberties groups, industry and federal and state law enforcement.

The White House account of the meeting had not appeared on the Briefings & Statements page of the White House website as of early Friday.

Why It Matters

The plan fills in part of a gap the task force left open last week. On Oct. 9 it said that disclosing AI incidents is not optional, but it named no receiving agency, deadline, legal authority or penalty. Thursday's outcome points toward a channel and a format for such reports. It still does not say whether developers would be required to use them or what happens if they do not.

For boards and executives, a federal early-warning system for frontier AI labs would change the flow of information about risks that reach their own organizations. Companies that build on models from these developers often learn of a vendor's security problems only through public disclosures or contract terms. A government channel could mean officials hear about an incident before customers do, which makes vendor notification clauses in AI contracts worth reviewing now.

The enforcement committee also matters. The FTC's chair, Andrew Ferguson, is a vice chair of the task force, and the agency has sent civil subpoenas to OpenAI and Anthropic over AI safety incidents, according to Axios. Pairing consumer protection enforcers with the Justice Department suggests the administration sees existing law, not new legislation, as its main lever.

Context

President Donald Trump created the task force Oct. 4 and put Clayton in charge. Clayton told The Wall Street Journal at the time that the group had 120 days to report on the technology's risks and opportunities.

Trump has rejected calls for more federal AI regulation, arguing it would hurt the U.S. in its competition with China, Bloomberg noted. Last month he brokered a voluntary agreement with tech executives under which outside auditors assess the safety of AI systems and company leaders meet regularly to set industry practices. The White House has given no plan for monitoring that pact.

AI has also become a contested issue ahead of the midterm elections, with public worry about job losses and loss of control over AI systems, Bloomberg reported.

What to Watch

The first marker is whether the White House publishes its account of the meeting and sets a deadline and scope for the advisory council's report. The 120-day window Clayton described would end in early February.

Next is the design of the lab reporting protocols: which agency receives the warnings, what counts as a reportable breach or jailbreak, how quickly labs must report and whether participation is voluntary. Also watch which agencies join the enforcement committee and whether the FTC's subpoenas to AI developers become its first test.