AI Slop Is Swamping a House Office That Drafts US Laws

Author: Owen Dahlkamp (Politico) URL: https://www.politico.com/news/2026/08/17/ai-slop-lawmakers-congress-01008376 One-line: The House Office of Legislative Counsel — the 61-attorney quality-control layer between a member’s idea and a bill — is being swamped by AI-generated drafts riddled with errors, and its inability to use commercially available AI in response creates a dangerous asymmetry.

Key claims

• Congressional staffers and outside advocacy groups are feeding legislative ideas directly into ChatGPT and Claude, generating full bill text that they then submit to the Office of Legislative Counsel (OLC) for review — “going to Claude or ChatGPT to draft the legislative language itself.”
• The OLC, which employs 61 attorneys and 19 support staff and prepared over 30,000 bills in the last Congress (nearly 3x the number actually introduced), spends “more time trying to fix AI-drafted legislation than it would take for them to draft it from scratch.”
• Specific error types: wrong terms of art (whether a payment is a “tax credit, tax deduction, tax exclusion or a grant”); incorrect citation of the U.S. Code — “a guaranteed way of introducing thousands of bugs into our legal system”; bills that accidentally exclude D.C. and tribal nations from federal programs by defining “state” as only the 50 states.
• House saw a 70% increase in bills proposed over the past decade; in early 2025, the first 60 days of Congress produced 5,623 legislation requests — up 72% from two years earlier.
• The OLC itself cannot use commercially available AI tools, creating “a complete incongruency”: advocacy groups get AI-speed drafting, the lawyers reviewing them work at human speed.
• The Comparative Print Suite — a natural-language-processing tool developed by Ari Hershowitz and OLC staff — already exists and allows congressional staffers to visualize how a bill would amend the U.S. Code. Designed specifically to return error messages when uncertain (“no hallucinations”), it’s the first AI tool adopted in the House. Senate adoption remains limited.
• AI-generated bill drafts also tend to be wordier, and staffers who use AI to draft legislation lose depth of subject-matter understanding — they’re no longer “forced to learn the issue as deeply.”
• Rep. Anna Paulina Luna (R-FL) used Claude to draft a committee amendment summary; she confirmed the amendment itself wasn’t AI-written but told staff to “double check and be more thorough.”

Notable quotes

“AI cannot replace the skilled and dedicated professionals in the Office of Legislative Counsel who help the House do its most critical job.” — Rep. Joe Morelle

“Relying too heavily on AI-generated bill text could introduce errors, create unintended legal consequences, or result in language that does not accurately reflect a member’s policy intent… That responsibility cannot be outsourced to a chatbot.” — Rep. Norma Torres

“AI is good for a lot of stuff. But it’s not capable of drafting legislation that you would want to enact into law.” — Daniel Schuman, American Governance Institute

“AI is going to miss those nuances.” — Wade Ballou, former OLC head (2016–2024), on the distinction between a tax credit, deduction, exclusion, or grant

“You had a problem that was already at 10. Now you multiply it by 100.” — Ari Hershowitz, on AI-accelerated legislative volume

“Members and interest groups get to write things with AI, but then leg counsel themselves can’t use [commercially available] AI. There’s a complete incongruency.” — Charlotte Lee, Granicus

“It’s absolutely terrifying, and no one is thinking about it yet.” — anonymous House staffer

How it connects

• Slopulism — this is AI slop applied to government: AI-generated content that looks like legislation but is plagued by the same hallucinations and attention failures that make AI bad at nuanced institutional tasks.
• AI Cheating in Education — parallel dynamic: using AI to skip the hard intellectual labor (drafting) that builds the competence needed to catch AI errors; the OLC story is the governance equivalent of the education cheating story.
• Artificial Intelligence — the underlying technology creating the asymmetry between AI-speed drafting and human-speed review.
• Governance — the OLC is “the institutional capacity that makes Congress’s primary constitutional power, lawmaking, actually work” (Marci Harris, POPVOX Foundation); AI pressure on it is a governance integrity issue.
• Tech Regulation — the piece is itself an artifact of the regulatory and governance conversation about how AI interacts with government institutions.