AI-Generated Bills Create More Work Than They Save, Overwhelm House Legal Counsel

AI-generated bills are reportedly causing problems in the Capitol - Engadget

They're so full of errors that Congressional lawyers are spending a lot of time trying to fix them, Politico says. The House Office of Legislative Counsel is swamped with AI-generated bills that are riddled with wrong terms, incorrect citations and other mistakes, according to Politico. More and more representatives' offices are using publicly available general AI tools like ChatGPT and Claude to write their legislative proposals. Politico's sources said lawyers at the US House Office of Legislative Counsel (OLC) are spending more time to review and rewrite them than they would have if they had written the bills from scratch. The sources explained that AI tools tend to miss nuances and tiny crucial details, which could have a huge effect on how a law is interpreted and enacted. Wade Ballou, who headed the office for almost 10 years until 2024, said AI can't determine whether a pot of money should be a "tax credit, tax deduction, tax exclusion or a grant," for instance. Sometimes, its classification of "state" only includes the 50 states, which would exclude DC and tribal nations from federal programs. A lawyer who worked with the office also told Politico that AI would incorrectly cite previous statutes in drafts. AI-generated proposals are also creating a different kind of problem: Congressional staffers reportedly aren't as deeply familiar with what their bills are about and hope to accomplish anymore. Because they use AI to draft their proposals, they're not forced to "learn the issue as deeply." In order to keep up with the increasing workload brought about by AI-drafted bills, the OLC is, well... exploring the use of AI to "improve efficiencies." A working group within the office has developed an AI tool called "Comparative Print Suite" that can help staffers visualize how a proposal will change current laws. Unlike general AI tools, it returns an error if it can't figure out where changes in the proposal should be made, preventing hallucinations from making it into draft bills.

AI-drafted bills take longer to fix than to write

Congressional staffers and outside groups are using Claude and ChatGPT to write legislation, Politico reports. The House Office of Legislative Counsel now spends longer fixing AI-drafted bills than it would spend writing them from scratch. Its own testimony says it is bracing for the workload. Congressional staffers and outside groups are using Claude and ChatGPT to draft legislation. The lawyers who check that work cannot keep up. Owen Dahlkamp reported the strain on the House Office of Legislative Counsel for Politico on Monday. He drew on interviews with eight current and former officials who work or have worked with the office. One person who has advised the office on the technology gave Politico the line that carries the story. The office spends "more time trying to fix AI-drafted legislation than it would take for them to draft it from scratch," the person said. Politico granted anonymity to several people because they lacked authorisation to speak on the record. Four congressional staffers involved in drafting told Politico that AI use is becoming routine in the House. "It's absolutely terrifying," one House staffer said, "and no one is thinking about it yet." What the office does The Office of Legislative Counsel is the quality-control layer between an idea and a statute. "When we talk about Congress's ability to legislate, we are talking about this office," Marci Harris of the POPVOX Foundation told Politico. The volume is already large. In the 118th Congress the office prepared 30,494 bills, 188 percent more than the 10,564 actually introduced. That comes from testimony in March by legislative counsel Warren Burke. The office also prepared 21,484 amendments, up 46 percent on the 114th Congress. Burke's figures also show who does that work. The office had 78 attorneys and 23 support staff in the first session of this Congress. Attorney numbers are up 59 percent since the 116th. Average experience has fallen 22 percent over the same stretch, and 58 percent of attorneys now have less than a decade of service. Where AI goes wrong in a bill The errors described are small and consequential. Wade Ballou led the office from 2016 to 2024. He told Politico that AI struggles to tell whether a pot of money should be a "tax credit, tax deduction, tax exclusion or a grant", and that AI will miss those nuances. Definitions cause similar trouble. A bill that defines a state as only the 50 states can exclude Washington DC and tribal nations from a federal programme. A person who works with the office gave Politico that example. Citations are the third failure. Ari Hershowitz is a lawyer and technologist who has worked with the office. He called incorrect statutory references "a guaranteed way of introducing thousands of bugs into our legal system". Daniel Schuman of the American Governance Institute put the general case. "AI is good for a lot of stuff," he told Politico. "But it's not capable of drafting legislation that you would want to enact into law." Official pipelines have buckled under machine-generated submissions before. In August we reported that someone filed 55 software vulnerabilities in days, and that 54 were AI-invented. Two different growth figures Politico cited a 72 percent rise in requests, drawn from Burke's testimony last year. That figure covers the first 60 days of this Congress. It compares 5,623 requests against 3,271 in the same window of the previous Congress. Burke's most recent testimony reports a different rate over a longer window. Across the first 13 months of this Congress the office received 23,239 requests. The comparable figure for the 118th was 21,979. That is a rise of 5.7 percent. On AI specifically, Burke was direct in March. "In the near term, we are bracing for AI to expand our workload," he wrote. He added that text produced by outside entities, human or computer, often takes significantly longer to review and revise. Burke declined to comment for Politico's article. The office is testing its own AI The office has organised a group of staff to evaluate Microsoft Copilot, as authorised by the House. The stated aim is improving efficiencies such as legal research, according to Burke's testimony. It already runs one tool of its own. The Comparative Print Suite uses natural language processing to show exactly how a bill would change the US Code. Hershowitz described it to Politico as the first AI tool in the House. It returns an error when it cannot place a change with confidence. "There's no hallucinations," he said. Charlotte Lee of Granicus helped design that tool. She described the wider arrangement to Politico. "Members and interest groups get to write things with AI, but then leg counsel themselves can't use [commercially available] AI," she said. "There's a complete incongruency." What has not been established Politico reported that no bill voted on by either chamber has turned out to be entirely AI-crafted. The evidence so far concerns AI-drafted bills arriving at the office, not statutes leaving Congress. Disclosure is patchy by nature. One staffer told Politico that people who use AI tend not to talk about it, and that it still feels taboo. Labelling rules elsewhere have landed unevenly too. Under the EU AI Act, as we reported this month, an AI-written article needs no label while a proofread email does. One incident is public. Representative Anna Paulina Luna drew criticism in June over an amendment summary generated with Claude, though she said she had not used AI to write the amendment itself. "Not a shocker. Most staff use it," she said, adding later that she had told her staff to double check. Mike Pearl of Gizmodo reported that she later deleted one post on the episode and replaced it. Two members of the Committee on House Administration gave statements. "AI cannot replace the skilled and dedicated professionals in the Office of Legislative Counsel," said Representative Joe Morelle. Representative Norma Torres said responsibility for introduced legislation rests with members and their staff, and "cannot be outsourced to a chatbot". The wider AI push on Capitol Hill Congress has been circling the technology from several directions this year. It spends nine of ten AI dollars with OpenAI. House Democrats want answers from OpenAI and Anthropic on rogue agents. Legislation is moving as well. A bill from Ted Lieu and Jerry Moran would create an emergency kill switch for AI systems. That followed a rogue agent breaching Hugging Face in July. Ballou sees a narrower role for the technology. AI can help lawmakers process information, he told Politico, but cannot replace the art of drafting. Hershowitz expects the volume problem to worsen either way. "You had a problem that was already at 10," he said. "Now you multiply it by 100."

'It's Absolutely Terrifying': AI Is Reportedly Slopping Up the Bills in Congress

Four anonymous congressional staffers who apparently help draft bills spoke to Politico for a story published Monday. They say AI-generated text is seeping into bills at an alarming rate -- alarming, because when they pass, bills turn into the federal statutes that are part of the bedrock of American life. "It's absolutely terrifying," one of them told that publication. If the report is true, AI in the legislative process sounds like it does something similar to AI in the lawsuit-writing process: gives more people confidence to try and generate documents in jargon-heavy language that needs to be precise, while also making more work for the people who eventually have to read those documents. Last year while addressing the House, Warren Burke, who leads congress' Office of Legislative Counsel -- a collection of mostly lawyers who format and draft laws, request for legislation received by the office were up 72% in 2025 compared to two years earlier. One of Politico's anonymous staffers said people and organizations outside congress are a big part of the issue. They're "going to Claude or ChatGPT to draft the legislative language itself," the person said. They added that the OLC spends "more time trying to fix AI-drafted legislation than it would take for them to draft it from scratch." In a puzzling and somewhat ambiguous incident in June, Representative Anna Paulina Luna, a Republican from Florida, posted apparent evidence online that Anthropic's Claude chatbot may have been used by a staffer to draft legislation. Later, she seemed to confirm that it had been used to at least edit a bill, but the tweet in which she said that was deleted and replaced with a tweet saying the chatbot was only used for a summary of the bill.