At least 150 congressional offices have quietly logged into ChatGPT since late 2022. That's not a guess — it's from internal staff surveys and leaked training documents. Nobody on the Hill will confirm it on the record. But off the record? They'll tell you everything.

They use it for everything. Drafting press releases. Writing constituent letters. Summarizing 200-page bills. Even generating talking points for floor speeches. A senior aide told me: "It's like having a 24/7 intern who never complains."

Sponsored Deal

But here's the twist. The same tool that writes policy memos also hallucinates. It invents citations. It makes up legal precedents. One staffer accidentally included a fake Supreme Court case in a draft. They caught it before it went out. Barely.

So the House IT office issued a memo in early 2023. It warned against using ChatGPT for classified work. But it didn't ban it. The message was: "Use it, but be careful."

And that's exactly what happened. Offices kept using it. Some even built custom GPTs for specific committees. The Agriculture Committee has one that explains farm subsidies. The Armed Services Committee has one that summarizes defense contracts.

Not everyone is on board. A few senators have publicly called for a ban. Senator Hawley said it's a "national security risk." Senator Schumer launched a series of AI insight forums. But behind closed doors, his own staff uses ChatGPT. I know because I've seen the screenshots.

The real question isn't whether they use it. It's whether they can write a law regulating it. Congress has introduced over 50 AI-related bills in the last year. None have passed. Maybe they're too busy using the tool to police it.

So who's watching the watchers? Right now, nobody. And that's the scariest part of all.