Overview
- Lawmakers and staff are widely using chatbots such as Claude, ChatGPT and Microsoft Copilot to draft speeches, constituent replies and legislative text, producing faster first drafts but frequent errors that require heavy revision.
- The House Office of Legislative Counsel is seeing a sharp rise in AI-written submissions and now spends more time rewriting flawed drafts than it would drafting bills from scratch, creating a downstream backlog in the law‑writing process.
- Rep. Anna Paulina Luna's office filed an amendment in June that contained unedited AI output, illustrating how raw chatbot text can enter the legislative pipeline and risk producing lasting legal ambiguities.
- Formal House and Senate guidance on staff use of AI exists but lacks centralized enforcement, so policing and training fall to individual offices and produce inconsistent rules and practices across Capitol Hill.
- New specialized tools such as the Republican Study Committee’s Byrd Bot are being used to test or generate language to meet procedural rules, and experts warn that widespread, unchecked use could embed persistent errors and erode Congress’s institutional drafting skills.