U.S. Republican Representative Anna Paulina Luna (안나 파울리나 루나) has been caught up in controversy after a draft amendment left traces of an artificial intelligence chatbot.

According to tech outlet Engadget on Wednesday, the wording at issue was “Claude responded:” left in the draft. That raised the possibility it was not cleaned up after content exchanged with Anthropic’s AI chatbot Claude was pasted in.

As the controversy grew, Luna acknowledged on X, formerly Twitter, that staff used AI while correcting wording in the draft. She said staff used AI to fix the draft text and did not finish editing it, adding that most staff use AI. The post was later deleted.

Luna then issued an additional explanation. In a later post, she claimed the staff used AI not on the amendment text itself but to check spelling and grammar in the amendment “summary.” She wrote, “No legislation is written by AI,” and said the screen in question was an AI summary of the bill and material for a spell-check.

The dispute is whether it was merely used for spell-checking or whether AI actually intervened in drafting the bill language. The phrase “Claude responded:” left in the draft can be read less as a trace of a standard proofreading function than as a marker created while moving chatbot dialogue into a document. Luna’s initial explanation said the draft text was proofread with AI, before she later narrowed it to say it was not the actual bill text.

In a subsequent post, she also pushed back by asking who raised the suspicion. But the office’s explanation changed several times in a short period. It first appeared to treat AI use itself as not a major issue, then drew a line by limiting it to a summary and spell-checking.

The controversy can be read as an example of how deeply generative AI has entered political document drafting and review processes. Luna’s comment that “most staff use AI” suggests AI tools are already widely used in staff work. But if chatbot traces are left intact in official documents such as draft bills, controversy over responsibility for drafting and review procedures could grow.

In particular, Luna’s final emphasis that “no legislation is written by AI” shows politics is trying to explain AI use separately from actual drafting responsibility. In this case, the wording left in the draft itself weakened that explanation. How far the chatbot was involved in drafting work, beyond spelling and grammar tools, remains at the center of the controversy.

Yeah my staff used AI to spell/grammar check the amendment SUMMARY, not the actual amendment text itself. Not a shocker. Most staff use it. I have told them to make sure they are double checking and more thorough. What dork planted this story? Btw love Claude but Grok is way more…