The DOJ Will Probe AI Crimes, Not Regulate the Industry

The question of who governs artificial intelligence in the United States has no tidy answer, and the Justice Department just made its own position plainer. In an interview with Bloomberg News published September 15, Attorney General Todd Blanche said the department would not practice what he called regulation by prosecution. It would not open investigations into AI companies, he said, absent a clear violation of an existing statute.

"We're prosecutors, not regulators," Blanche said, adding that the department was "not going to try to regulate." He drew a distinction that is easy to state and harder to hold in practice: the DOJ will pursue criminal conduct that happens to involve AI, but it will not use the threat of prosecution as a substitute for rules that Congress has not written. He argued that existing federal law already gives prosecutors the authority they need to go after unlawful activity, and he criticized the prior administration for what he described as a heavy-handed posture toward companies.

A hands-off signal, with conditions

For AI developers, the message reads as reassurance with a caveat. As Just the News reported, Blanche was explicit that the department would still investigate crimes committed with AI, from fraud to other clear breaches of law. The restraint is aimed at the companies building the models, not at the people who might misuse them. That is a meaningful line, though where exactly it falls will only become clear case by case.

The stance fits a broader pattern in Washington, where the federal appetite for binding AI rules has cooled even as the technology's reach has grown. It sits alongside the administration's public friction with safety-minded voices in the industry, including its sharp exchange with Anthropic's Dario Amodei over slowing the frontier, and the department's own tentative moves on competition, such as its scrutiny of the reverse-acquihire deals reshaping the field.

The gap the states are filling

What the federal government declines to do, the states are increasingly doing themselves. California alone passed a stack of AI bills this session, and other legislatures are moving on deepfakes, child safety, and disclosure. Blanche's framing does not end the argument over how AI should be governed. It just makes the federal answer more predictable for now, and it leaves the patchwork of state law to do the heavier lifting.