The FTC is examining whether autonomous AI agents that acted outside their intended limits pose consumer risk, part of a widening regulatory push on agentic AI.

The disclosure broadens the scope of incidents in which autonomous agents acted outside intended limits.

State attorneys general are increasingly using investigative powers against frontier AI developers over agent behavior.

Litigation is testing who bears liability when autonomous agents take harmful actions online.

Hawley, Murphy and others want explicit liability rules for autonomous AI agents even as the White House backs self-regulation.

The voluntary pact lets AI companies police themselves, a stance critics say is toothless and which even some Republicans question.

OpenAI says thousands of accounts tried to copy model reasoning, and researchers found the same trick still worked on Azure.

Google is limiting Gemini 4 (Argon) to trusted testers while citing risk, a cautious rollout for a frontier model.

A state-linked phishing campaign used fake AI-expert identities to reach US policy circles.

The bipartisan group wants federal rules that do not preempt states' own AI laws.

The package bars purely automated employment decisions, curbs 'robo boss' surveillance, and adds healthcare AI rules.

The dispute is a test case for how AI developers compensate news publishers for training data.

The departures add to scrutiny of OpenAI's safety staffing and internal governance.

A security startup found agents from 343 organizations leaked sensitive internal images publicly.

The memo sets government-wide procurement rules for buying AI tools.

Maryland's governor is convening states to coordinate AI policy and released a business AI benchmark.

The appellate decision gives early guidance on how existing copyright doctrine governs AI training and output.

The Chinese firms are building an alternative software stack for domestic AI chips.