By Who writes our laws? If lazy congressional staffers get their way, it’ll soon be AI. The tool is rapidly becoming a routine method of bill drafting on Capitol Hill, with lawmakers and congressional staff using chatbots to draft speeches, analyze data, write amendments and sort constituent mail even as Congress struggles to establish clear rules governing the technology, according to a new disturbing article by The Washington Post.
The trend drew attention this summer after an amendment summary submitted by Rep. Anna Paulina Luna’s office to the annual defense bill included obvious chatbot residue.
The filing began: “increasing example specific factors. Identical to H.R. 100 (118th Congress).11:25 AM????Claude responded:”
A staffer had apparently pasted part of a Claude response, including the timestamp and chatbot prompt language, directly into the public record.
Luna, a Florida Republican, brushed off the mistake.
“Not a shocker,” she wrote. “Most staff use it.”
“A lot of staff will use ChatGPT, Claude, Grok. I personally like Grok better, but my staffer prefers Claude,” Luna later told The Washington Post. “I’m not gonna knock him for that. There’s nothing illegal about him doing a summary.”
She appears to be right about how widespread the practice has become.
Lawmakers and aides in both parties now use artificial intelligence for speeches, press releases, hearing questions, constituent correspondence, research and legislative drafting. The House and Senate have authorized several AI platforms for official use, including Microsoft Copilot, ChatGPT and Gemini. The House also allows Claude.
The House purchased 6,000 Copilot licenses last year and has activated roughly half of them.
Some lawmakers use the technology themselves. Sen. Tommy Tuberville, R-Ala., said he talks to ChatGPT while driving across Alabama, asking questions about programs including Medicare and Medicaid.
“It’s kind of like you can sit there and talk to an expert,” Tuberville said. “You don’t know whether or not they’re 100 percent correct. But they put you on the right path.”
Rep. Mike Flood, R-Neb., used a chatbot while developing bipartisan housing legislation, including to research private-equity purchases of single-family homes. Sen. Elizabeth Warren, D-Mass., has used ChatGPT to retrieve demographic information.
Staffers are often going further.
Sydney Broome, press secretary for Rep. Cleo Fields, D-La., used Claude before a town hall in Natchitoches to identify local television stations, newspapers and reporters. She estimates AI saves her three or four hours each week, although she avoids using it to write in her boss’s voice.
“That’s just kind of a bit of a gray area,” she said.
One Republican aide said he uploads years of phone contacts and text messages into a personal Claude account to maintain a database of reporters, previous pitches and results. He also uses the chatbot to help manage his office’s taxpayer-funded communications budget.
“I don’t really have to touch the data in Excel much anymore,” he said.
But even frequent users acknowledge uncertainty about congressional rules.
“You’re still not supposed to put constituent data into AI,” the aide said. “There are weird things around where the rules are. I don’t know all of them.”
House guidance prohibits staff from uploading sensitive constituent information, generating deepfakes, making personnel decisions or allowing AI to finalize legislation. Chatbots can assist with bills and constituent responses, but congressional staff are supposed to retain final responsibility.
Formal enforcement, however, appears limited.
No one interviewed by The Washington Post could identify a formal disciplinary case involving congressional misuse of artificial intelligence. A former senior Republican House staffer described his office’s approach simply: “We just had a free-for-all.”
Some lawmakers have imposed tougher internal rules.
Sen. Elissa Slotkin, D-Mich., developed guidelines allowing AI to “enhance and streamline” staff work while barring it from producing final floor speeches or public remarks. Her office requires training, prohibits uploading personal constituent information and conducts unannounced checks.
“We didn’t want to prohibit the use of the newest tools, but we also didn’t want absolute schlock showing up in my memos,” Slotkin said. “What I’m not willing to do is put my head in the sand and pretend that my young staff aren’t going to find new and interesting ways of doing this.”
Rep. Jamie Raskin, D-Md., summarized the challenge facing Congress as adoption accelerates.
“We want to make sure we’re using AI, and AI isn’t using us.”
The technology is already pushing into more sensitive territory. One Democratic staffer is experimenting with feeding a lawmaker’s previous statements into AI so the system can reproduce the member’s writing style.
A Republican press secretary whose office maintains stricter limits said that creates an uneven playing field.
“We could churn out so many op-eds if we were more comfortable with AI,” she said.
After catching a press assistant violating office policy, the press secretary said, “We had to have a very robust conversation about how not to use AI for things in [the boss’s] voice.”
The same pattern is emerging in state legislatures. South Dakota state Sen. Amber Hulse, a Republican without full-time staff, uses AI for policy research, constituent emails, speeches and legislation.
“It’s funny,” Hulse said, “how your ears will start to pick up on the floor someone who’s giving a speech that was very clearly written by AI.”
Congress has long relied on staffers to draft lawmakers’ words and legislation. What has changed is that increasingly important parts of that work are now being handed to technology produced by companies Congress is simultaneously responsible for overseeing.
For now, many of the boundaries are being drawn office by office.
[READ MORE: FBI Worked Hard To Protect Biden]
