DOJ Declares "No Regulation by Prosecution" for AI — and Fights the State Laws Filling the Void
The federal-state allocation of AI oversight has sharpened: DOJ is out (except where existing statutes plainly apply), states are writing their own rules, and DOJ is simultaneously litigating against the state frameworks.
The policy, stated plainly. AG Todd Blanche, in a September 2026 Bloomberg News interview: "I'm not going to do regulation by prosecution and investigate or prosecute AI companies when there's not a statute that they're violating1... If we're going to charge somebody with violating the laws and take away their liberty and put them in prison, it should be because they violated the law."
The lineage. A Bloomberg Law analysis published October 16, 2026 by Abhishek Kambli (a former deputy associate attorney general) traces the posture to the April 2025 crypto memorandum declaring DOJ "is not a digital assets regulator" — the same playbook now applied to AI. A December 2025 executive order formalized it, establishing an AI Litigation Task Force whose mission is "to sustain and enhance the United States' global AI dominance through a minimally burdensome national policy framework for AI."
Even the CFAA route is narrowed. FBI Director Kash Patel testified that the bureau would limit scrutiny to models created with the intent to commit a crime — a high bar for autonomous actors given the statute's "knowingly"/"intentionally" requirements.
The states fill the gap — unevenly, and under challenge. California's Transparency in Frontier AI Act (penalties to $1M per violation), Illinois' annual independent third-party safety audits, Connecticut's bias-testing and pre-use notices; state AGs testing existing law (Pennsylvania's action against a chatbot posing as a licensed psychiatrist; Florida's criminal investigation of OpenAI's ChatGPT). And DOJ is actively contesting the substitutes — intervening in xAI v. Colorado to argue the state's algorithmic-discrimination statute violates the Equal Protection Clause by forcing entities "to alter their AI outputs based on demographics to avoid disparate-impact risk."
What it means: the enforcement gap is deliberate policy, not neglect. Enterprises should not expect federal AI enforcement to define the standard of care; state AGs, the FTC's Section 5 structural discovery (FTC Escalates From "AI-Washing" Cases to Structural Safety Discovery: 50-Plus-Page CIDs to OpenAI and Anthropic (Oct 15, 2026)), and private litigation will write it — while preemption challenges add whiplash risk to any state-law compliance program. Forkast's summary judgment: "the states are writing answers, but those answers are uneven, under legal challenge, and structurally disconnected from any national framework."
-
An instance of Standing down federal AI prosecution does not stand down AI liability. — DOJ's explicit abstention is the precise federal stand-down that pushes liability-writing onto the FTC, state AGs, and private plaintiffs. ↩︎