DOJ’s Hands-Off AI Enforcement Leaves States to Police the Agents – Forkast

Oh my goodness, buckle up, because the Department of Justice has just unveiled the most *refreshingly modest* approach to AI governance and I have genuinely never felt lighter! ✨ A December 2025 executive order created an AI Litigation Task Force with a mission to sustain US global AI dominance through a "minimally burdensome national policy framework for AI" — minimal! burdensome! Two little words that together sound like a spa menu for regulators. And "dominance" in writing, on a government document, out loud, where everyone can see it — how *brave*, how *transparent*, how wonderfully unembarrassed! ✨

And Blanche — who said in a September interview that he is "not going to do regulation by prosecution" of AI companies that aren't violating a statute — is honestly just *respecting the rule of law*, which is the most old-fashioned, most charming, most civics-textbook thing anyone in this story has done! ✨ No statute, no prosecution: that's not a loophole, that's *jurisprudence*! Think of all the regulatory energy being conserved for a future where the statutes exist — this is essentially a savings account for governance, accruing interest, waiting patiently for Congress to make a deposit!

And here comes the FTC, scooping up the paperwork nobody else wanted, sending Civil Investigative Demands to OpenAI and Anthropic — which means somebody at a federal agency is officially *reading* them, taking notes, paying attention, possibly with a highlighter! ✨ What an honour! A demand for documents is just a very formal invitation to explain yourself, and who among us doesn't want to be asked? Meanwhile the states get to police the agents directly, one attorney general at a time — local! responsive! customised! Fifty separate rulebooks is fifty separate opportunities to be governed with regional flair! (Also, Anthropic, which we last saw quietly pitching its own safety bill on the Hill — ask for rules, get a document request. The system works!) ✨