Rep. Luna denies AI wrote defense amendment after Claude appears in summary

Rep. Anna Paulina Luna (R-FL) found herself defending her office's AI use after screenshots circulated showing what appeared to be Claude-generated text in an amendment summary for the 2027 National Defense Authorization Act. The screenshot included the line "Claude responded:" followed by summary text. Luna initially posted that "staff used AI to correct a draft text and didn’t edit," adding that most staff use it and she told them to be more thorough. After speculation grew, she edited her response to clarify that AI was only used for "spell/grammar check the amendment SUMMARY, not the actual amendment text itself." She further insisted that "NO Legislation is ever drafted with AI" and that bill text comes from the House Legislative Council, which is prohibited from using AI.

The incident highlights the growing tension around AI use in government drafting. Over the past few years, judges have caught lawyers using AI chatbots to generate filings with fake citations. In Brazil, city officials unknowingly approved an ordinance written by ChatGPT. Arizona state representative Alexander Kolodin has admitted to using ChatGPT for state-level legislation. Luna's case adds to the scrutiny, with her initial and edited responses revealing the difficulty of drawing clear lines between AI-assisted wording and actual legislative drafting. As AI tools become ubiquitous in professional settings, lawmakers face increasing pressure to disclose when and how they rely on them for official documents.