AI tools are increasingly used by congressional staff in both the House and Senate to draft and research materials, including legislative text, with little formal oversight or discipline for misuse.

This trend is framed as part of a broader government push to rapidly integrate AI across agencies, under a Trump administration policy stance, leaving regulations and accountability largely unresolved.

A notable incident saw Representative Anna Paulina Luna’s office accidentally inserting a Claude prompt output into the public record of the National Defense Authorization Act, illustrating how AI outputs can slip into official documents.

Overall, the piece highlights growing reliance on AI in lawmaking, the risks of unchecked use, and the expectation that federal regulatory frameworks may remain ineffective in addressing these challenges for the foreseeable future.

The lack of concrete oversight on Capitol Hill means AI-generated content could introduce nonsensical language or hallucinations into speeches, emails, and laws without detection.

Staffers acknowledge using tools like ChatGPT, Claude, and Grok, with some preferring one tool over another, and a few, such as Sydney Broome, adopting self-imposed safeguards (e.g., not writing documents from scratch with AI) despite no formal rules on AI usage in Congress.