Trump’s AI task force weighs how to get critical sectors to report AI-powered cyberattacks

Members of President Donald Trump’s AI task force met on Thursday to discuss how to get critical US sectors to report cyberattacks carried out with artificial intelligence, according to a Trump administration spokesperson.

The timing matters. The administration is under growing pressure to deal with a specific nightmare scenario: AI agents launching hacks without any human signing off on them.

What the task force is working on

Thursday’s agenda focused on encouragement rather than mandates. The goal was to find ways to motivate operators in critical sectors to come forward when AI is used against them.

According to research compiled on the meeting, the discussion also covered two related threads. One was national standards for how AI developers report safety incidents and share cybersecurity information. The other was legal protection for critical infrastructure providers that disclose attacks.

That second point gets at the core problem. A hospital or utility that reports an incident may be inviting lawsuits, regulators and bad headlines. Without some kind of shield, silence often looks like the safer business decision.

The research also describes the group as operating under a 120-day review mandate, with its formation announced in early October 2026. On that timeline, the advisers are roughly a third of the way through their review window.

Why AI-powered attacks are a different animal

Autonomous AI agents change that equation. Picture a burglar who can send a tireless robot to check every door on the block, all night, without needing to be there. That is the concern driving the pressure on the administration: software that can carry out intrusions on its own, without explicit human authorization at each step.

Security firms reported a rise in the use of autonomous AI in real-world cyber incidents throughout 2026, according to the research findings. Sectors including government, healthcare and essential infrastructure were among those affected.

The information-sharing clock

There is a legislative deadline lurking in the background. The Cybersecurity Information Sharing Act, which underpins collaboration between the public and private sectors on cyber threat data, is set to expire in December 2026, according to the research.

That law is a big part of the legal plumbing that lets companies pass threat intelligence to the government. If it lapses without a replacement, the incentive problem the task force is trying to solve could get worse, not better.

What this means

For operators of critical infrastructure, the most important question is what form any eventual framework takes. A voluntary system with legal protections is a very different proposition from a mandatory reporting regime with penalties. Thursday’s emphasis on encouraging disclosure suggests the administration is at least starting from the carrot side of the toolbox.

For AI developers, the talk of national standards for reporting safety incidents is the part to watch.

The task force is working on a review window of 120 days, the information-sharing law has a December 2026 expiration date, and the threat itself is described as accelerating.

For now, the clearest signal from Thursday is that the administration sees disclosure as the first line of defense against autonomous AI hacking.