Fact Check: Did Anna Paulina Luna admit AI wrote her defense bill amendment before deleting it?

WASHINGTON, DC: Rep Anna Paulina Luna has come under scrutiny after an apparent artificial intelligence phrase appeared in an amendment summary submitted as part of the 2027 National Defense Authorization Act (NDAA).

The controversy quickly escalated when reports claimed Luna initially admitted her office used AI before deleting the post and issuing a narrower explanation.

This fact check examines what is confirmed, what remains unverified, and whether Luna actually reversed her position.

Claim: Anna Paulina Luna admitted AI wrote her amendment before deleting the post

Viral posts claim Luna initially admitted that her staff used artificial intelligence to write or edit the amendment, then deleted the admission after backlash.

The claim is based primarily on a deleted post that Gizmodo quoted as reading, "Yeah my staff used AI to correct a draft text and didn't edit."

According to the viral narrative, Luna later changed her explanation, denying that AI played any role in drafting legislation, which suggested an attempt to walk back an earlier admission.

Fact Check: AI artifact confirmed but deleted post cannot be independently verified

It is confirmed that an official amendment summary posted on the House Rules Committee website briefly contained the phrase "11:25 AM???? Claude responded," indicating an apparent output from Anthropic's Claude AI assistant. 

That summary was subsequently corrected, and multiple news organizations documented the editing artifact.

However, the exact wording of Luna's alleged deleted first post cannot currently be independently verified.

Gizmodo is the only publication that has reproduced and quoted the full text of the deleted message. No independently archived screenshot or public record confirming the precise wording has surfaced, meaning the reported quote remains single-sourced.

Luna's surviving public statement says her staff used AI only to "spell/grammar check the amendment SUMMARY, not the actual amendment text itself."

That explanation is more specific than the wording attributed to the deleted post but does not necessarily contradict it.

If Gizmodo's quotation is accurate, describing AI as being used to "correct a draft text" could reasonably refer to editing the amendment summary rather than drafting legislative language itself.

When later questioned by reporters, Luna maintained that the House Legislative Counsel drafts all official legislative text and that congressional rules prohibit lawmakers from submitting AI-generated legislation. 

She said the controversy involved only the amendment summary and added that she had jokingly criticized her own staff for using Claude to summarize the document.

The controversy gained additional attention because the NDAA includes provisions addressing artificial intelligence and military oversight, making the AI-generated summary particularly notable.

While the presence of an AI artifact in the summary is undisputed, no evidence has emerged showing that AI generated the official amendment language itself.