Systems & AdoptionNovember 13, 2026·4 min read

The DOJ Just Launched an AI Enforcement Unit. Google Just Made Opting Out of AI Impossible. Here's What to Audit This Week.

The DOJ created a federal unit to investigate unlawful AI use the same week Google tied Search access to AI adoption in schools. Shadow AI just became something an investigator can ask about.

By Forge Team · Examples are illustrative composites

If nobody at your company could hand over a complete list of the AI tools employees actually use — not the ones IT approved, the ones people actually open every day — that gap used to be an awkward silence in a meeting. As of this month, it's the kind of question a federal investigator can ask.

What happened

On November 5, the Department of Justice launched the Super Intelligence Task Force, a dedicated unit for investigating unlawful AI use, including AI-related fraud and corporate misuse. It isn't a working group drafting guidelines — it's an enforcement body that connects to the White House's Super Intelligence Force, which has a 120-day deadline for a comprehensive AI risk report. In the same week, Google told schools (Nov 3) that opting out of its AI tools now means losing Search access too, because the company treats the two as "deeply integrated." Former FTC chair Lina Khan called the bundling a form of monopoly leverage. And Anthropic's own evaluator, METR, estimated (Nov 5) that frontier model capability is improving at roughly 1.9 times its previous pace — meaning the gap between what a team has written down about its AI use and what AI can actually do is widening faster than most governance processes can track.

Put together, these are three versions of the same shift: the tools are getting more capable, the ability to avoid them is shrinking, and for the first time there's a federal body built specifically to ask what your team did with them.

What to do differently Monday morning

The question worth answering this week isn't "do we have an AI policy." A sentence in the employee handbook gesturing at responsible use doesn't answer the harder question underneath it: if someone asked you to name every AI tool a person on your team is using — including the free Chrome extension someone installed to summarize emails, or the voice-to-text app a salesperson uses for call notes — could you? Not eventually, after checking around. Today.

That's a shadow AI audit, not a policy rewrite. It takes an afternoon, not a committee.

An HR director at a 120-person regional bank

Priya ran what she thought was a straightforward exercise: ask each department head to list their team's approved AI tools. The lists came back clean — three tools, all vetted, all documented. Then she asked a follow-up question nobody had thought to ask: "what about the tools nobody approved?" Two recruiters were running resume screening through a free AI tool that had never gone through procurement, because it was faster than the bank's licensed system and nobody had told them not to.

Priya didn't ban the tool outright — she didn't have evidence it was unsafe, only evidence nobody had checked. She gave the recruiters one week to either get the tool reviewed or switch back, and added one line to the onboarding checklist: new hires list every AI tool they currently use, approved or not, before their first week ends. The point wasn't catching people. It was making sure the bank could answer the question before a regulator asked it first.

A marketing director at a 45-person SaaS company

Devon's team runs most of its workflow through Google Workspace, and the Search-AI bundling story landed differently for him: his team can't meaningfully opt out of Google's AI features without breaking tools they use for actual work, which means the decision about whether AI touches client data was effectively made by a vendor, not by his team.

He didn't try to reverse that. He did the thing within his control: he wrote down, for each Google AI feature his team actually uses, what data it can see and who on the team knew that before this week. Three people didn't. That's now fixed, and it took less time than the meeting where he found out.

The one thing

A federal enforcement unit changes the cost of not knowing what your team uses, and vendor bundling is quietly shrinking the number of places a "we never approved this" defense actually holds up. Spend one afternoon this week finding out what you don't currently know, before someone outside the company asks you first.

Reading about it only gets you so far

Patin turns this into five-minute drills that score what you write and tell you why. It's in closed beta — join the waitlist and we'll email you when your cohort opens.

Just want the writing? .