Rep. Anna Paulina Luna (R-FL) is pushing back on claims that her office used AI to write legislative text for a major defense bill. She says the technology was used for a narrow purpose: spellcheck and grammar review in an amendment summary.
The dispute began after accounts on X circulated screenshots tied to an amendment summary for the 2027 National Defense Authorization Act. The visible text appeared to include a reference to Claude, prompting users to question whether AI had been used beyond proofreading.
What the screenshot showed
The shared screenshot centered on an amendment summary, not the amendment text itself. According to the source article, the summary included the line: “Identical to H.R. 100 (118th Congress).11:25 AM????Claude responded: Requires the Secretary of Defense to designate Department of Defense activities, support, and operations at the southwest land border as a named operation with…”
That wording mattered because it appeared to show chatbot output inside a legislative workflow. The phrase “Claude responded” suggested that an AI tool had been involved, while the surrounding language referred to a defense-related amendment summary.
For readers following AI in legislation, the distinction between a summary and actual bill text is central. A summary is a description of what an amendment does. The amendment text is the legislative language that would be considered through the House process.
Luna’s explanation changed after speculation spread
Luna’s first response appeared to leave room for a broader reading. Her post said that “staff used AI to correct a draft text and didn’t edit.” She also wrote: “Not a shocker. Most staff use it. I have told them to make sure they are double checking and more thorough.”
After users on X began speculating that her staff was using the technology to write bills, Luna edited her response. The updated version narrowed the claim to the amendment summary.
“Yeah my staff used AI to spell/grammar check the amendment SUMMARY, not the actual amendment text itself,”
That clarification reframed the incident as a staff workflow issue rather than an AI-written bill allegation. Luna’s account was that the tool was used to check an explanatory summary and not to draft legislation.
House bill text was the line she emphasized
Luna followed with a second post that drew a hard boundary around legislative drafting. She wrote:
“FYI NO Legislation is ever drafted with AI. All bill text from the House comes from the House Legislative Council which is prohibited from using AI. The screenshot you’re referencing is an AI summary of the bill that’s also used for spellcheck, cmon man 🤣”.
That statement made two claims. First, Luna said House bill text comes from the House Legislative Council. Second, she said the House Legislative Council is prohibited from using AI.
The source article does not describe the exact workflow behind the screenshot beyond Luna’s explanation. It does, however, make clear that the public debate turned on whether AI touched a summary, the amendment text, or both.
Why AI references in official work draw attention
The incident fits a wider pattern described in the source article: as AI tools become more common in workplaces, references to AI chatbots are also showing up in places where they should not appear. The issue is not only whether an AI system was used, but whether its output was reviewed, removed, or properly separated from official material.
Legal and government settings are especially sensitive because mistakes can travel far. The source article notes that judges have caught lawyers using AI chatbots to draft legal filings filled with fake citations. It also points to lawmakers around the globe turning to the technology in official work.
Two examples from the source underline the range of AI use in public institutions:
- City officials in Brazil unknowingly approved an ordinance written with ChatGPT.
- Arizona state representative Alexander Kolodin told The Verge that he has used ChatGPT to write state-level legislation.
Those examples do not prove what happened in Luna’s office. They do show why a stray chatbot reference in a defense amendment summary can quickly become a larger question about process, review, and accountability.
The practical issue is verification
Luna’s defense rests on a narrow claim: AI was used for “spellcheck” and grammar in an amendment summary, while the actual amendment text was not drafted with AI. That distinction may be clear inside an office workflow, but it can become harder for the public to parse when a screenshot appears to show chatbot language embedded in official-facing material.
For congressional staff and other public officials, the lesson from the episode is straightforward from the facts presented: AI-assisted text can create confusion even when it is used for a limited task. A visible chatbot marker, a copied response, or an unclear summary can invite questions about who wrote what.
The broader debate over AI in legislation is not going away. Luna’s comments show one approach: permit staff use for checking a summary, while insisting that House bill text is handled through the House Legislative Council and not drafted with AI. The controversy shows how quickly that boundary can be questioned once AI output appears in the document trail.