Rep. Luna clarifies AI use in defense bill amendment, denies authoring legislation with chatbots
Florida congresswoman says staff used Claude for grammar-checking an amendment summary, not bill text, amid growing scrutiny of AI in legislative drafting.
Last verified:
Rep. Luna responds to AI use allegations in defense amendment
Rep. Anna Paulina Luna (R-FL) issued clarifications on June 24 regarding the use of Anthropic’s Claude in an amendment to the 2027 National Defense Authorization Act after X users circulated screenshots showing the AI tool’s involvement. According to The Verge, Luna’s initial statement acknowledged that her staff used AI to “correct a draft text,” which triggered immediate speculation that the bill itself was authored by the chatbot. She subsequently issued a second statement narrowing the scope of AI involvement to grammar and spell-checking of the amendment summary document only.
The amendment summary disclosure
The original screenshot showed an amendment summary for H.R. 100 with a visible Claude response embedded in the text, detailing requirements for the Secretary of Defense to designate border operations. The Verge reports that Luna’s staff apparently included the AI tool’s response within the document without fully removing or cleaning up the metadata. This created the appearance that Claude had authored substantive legislative language, though Luna has since denied this interpretation.
Legislative boundaries on AI use
According to The Verge, Luna emphasized that “NO Legislation is ever drafted with AI” and noted that “All bill text from the House comes from the House Legislative Council which is prohibited from using AI.” This distinction is material: the House Legislative Council operates under restrictions that bar automated drafting tools, while amendment summaries—which serve explanatory and administrative functions—are treated differently. Luna’s clarification suggests her staff used Claude to polish a summary document before submission, a practice she characterized as routine workplace technology use.
Broader context of AI in legislative drafting
The incident illustrates a widening pattern of AI tool integration in government workflows that has raised compliance and transparency questions. The Verge notes that Arizona state representative Alexander Kolodin has publicly disclosed using ChatGPT to draft state legislation, and city officials in Brazil unknowingly approved an ordinance written entirely with ChatGPT. Legal precedent shows similar risks: judges have caught attorneys submitting court filings with fabricated citations generated by AI chatbots, establishing a cautionary baseline for how automated tools can introduce errors into official documents.
Why This Matters
Luna’s case highlights the semantic boundary between permissible AI-assisted administrative work (grammar-checking summaries) and prohibited legislative authorship (drafting bill text). As AI tools become embedded in office workflows, the distinction between using them for clerical tasks versus substantive decision-making will become increasingly difficult to enforce and audit. For lawmakers concerned about AI transparency in government, this incident underscores the need for clearer policies defining where chatbots can and cannot be used in the legislative process—particularly given that staff may use these tools routinely without awareness of institutional restrictions or disclosure requirements.
Frequently Asked Questions
Did Rep. Luna's office use AI to write the actual bill text?
No. Luna stated that bill text comes exclusively from the House Legislative Council, which is prohibited from using AI. The AI involvement was limited to grammar-checking an amendment summary.
Why did Luna's initial response cause confusion?
Her first statement said staff 'used AI to correct a draft text and didn't edit,' which was ambiguous. She clarified after X users questioned whether the legislation itself was AI-drafted, explaining the AI tool was only used on the summary document.
Is this the first time a lawmaker has been caught using AI for legislation?
No. Arizona state representative Alexander Kolodin has publicly stated he used ChatGPT to write state-level legislation, and city officials in Brazil unknowingly approved an ordinance written with ChatGPT.