Blab AI
Rep. Anna Paulina Luna Defends Use of Claude AI in NDAA Amendment Summary
A curious phrase in a House Rules Committee amendment summary sparked a debate over artificial‑intelligence tools in Congress.
On June 24, 2026, a screenshot of the committee’s website showed the words “Claude responded” embedded in the middle of a description for a 2026 National Defense Authorization Act (NDAA) amendment. The insertion raised immediate questions about whether an AI assistant had helped craft the text.
The amendment, a summary of a change to the NDAA, was first posted to the committee’s page. According to the New Republic, the phrase “Claude responded” signaled that the assistant was involved. The description was later revised on the site, and the reference was removed.
Representative Anna Paulina Luna—Florida’s Republican freshman who introduced the amendment—addressed the controversy on X. She explained that her staff used Anthropic’s Claude chatbot to spell‑check and grammar‑check the summary, not to write the amendment itself. In a tweet she wrote, “Not a shocker. Most staff use it. I have told them to make sure they are double‑checking and more thorough.” She also asked, “What dork planted this story?”
The incident exposes a gap in congressional policy: the House Rules Committee, which governs how bills are presented, has no explicit rule on AI assistance. The use of Claude in this context is part of a broader trend of legislators employing language models for drafting support, a practice that has grown since Claude’s launch in March 2023.
Luna, born May 6, 1989, is a former Air National Guard member who has served in the House since 2023. She is the first Mexican‑American woman elected to Congress from Florida and sits on the House Committees on Foreign Affairs and Oversight and Government Reform.
Claude is a series of large language models built by Anthropic. Trained with a technique called “constitutional AI,” the model aims to improve ethical and legal compliance. Since its debut, Claude has been used in software development and increasingly in legislative drafting.
The NDAA authorizes the Department of Defense’s budget and policy for the fiscal year and is routinely amended by members of the House and Senate Armed Services Committees. Amendments are typically posted on the House Rules Committee website for public record.
No official investigation has been announced. The controversy has prompted lawmakers and policy analysts to call for clearer guidelines on AI use in the legislative process. Some commentators have warned that a lack of regulation could raise concerns about transparency, accountability, and the integrity of legislative documents.
At present, the only confirmed fact is that Luna’s staff used Claude for spell‑checking and grammar‑checking the amendment summary. The amendment description was subsequently edited to remove the phrase that identified the AI, and Luna’s statements on X confirm that the tool was not used to draft the amendment text.
The incident underscores the growing presence of AI tools in government workflows and the emerging debate over how to manage their use. While the House has not yet adopted formal rules, the controversy may prompt future policy discussions aimed at ensuring transparency and maintaining the integrity of congressional documents.
As of now, the amendment remains part of the 2026 NDAA, and no further action has been taken by congressional committees. The situation remains a point of reference for lawmakers considering how to regulate AI assistance in the legislative process.
On June 24, 2026, a screenshot of the committee’s website showed the words “Claude responded” embedded in the middle of a description for a 2026 National Defense Authorization Act (NDAA) amendment. The insertion raised immediate questions about whether an AI assistant had helped craft the text.
The amendment, a summary of a change to the NDAA, was first posted to the committee’s page. According to the New Republic, the phrase “Claude responded” signaled that the assistant was involved. The description was later revised on the site, and the reference was removed.
Representative Anna Paulina Luna—Florida’s Republican freshman who introduced the amendment—addressed the controversy on X. She explained that her staff used Anthropic’s Claude chatbot to spell‑check and grammar‑check the summary, not to write the amendment itself. In a tweet she wrote, “Not a shocker. Most staff use it. I have told them to make sure they are double‑checking and more thorough.” She also asked, “What dork planted this story?”
The incident exposes a gap in congressional policy: the House Rules Committee, which governs how bills are presented, has no explicit rule on AI assistance. The use of Claude in this context is part of a broader trend of legislators employing language models for drafting support, a practice that has grown since Claude’s launch in March 2023.
Luna, born May 6, 1989, is a former Air National Guard member who has served in the House since 2023. She is the first Mexican‑American woman elected to Congress from Florida and sits on the House Committees on Foreign Affairs and Oversight and Government Reform.
Claude is a series of large language models built by Anthropic. Trained with a technique called “constitutional AI,” the model aims to improve ethical and legal compliance. Since its debut, Claude has been used in software development and increasingly in legislative drafting.
The NDAA authorizes the Department of Defense’s budget and policy for the fiscal year and is routinely amended by members of the House and Senate Armed Services Committees. Amendments are typically posted on the House Rules Committee website for public record.
No official investigation has been announced. The controversy has prompted lawmakers and policy analysts to call for clearer guidelines on AI use in the legislative process. Some commentators have warned that a lack of regulation could raise concerns about transparency, accountability, and the integrity of legislative documents.
At present, the only confirmed fact is that Luna’s staff used Claude for spell‑checking and grammar‑checking the amendment summary. The amendment description was subsequently edited to remove the phrase that identified the AI, and Luna’s statements on X confirm that the tool was not used to draft the amendment text.
The incident underscores the growing presence of AI tools in government workflows and the emerging debate over how to manage their use. While the House has not yet adopted formal rules, the controversy may prompt future policy discussions aimed at ensuring transparency and maintaining the integrity of congressional documents.
As of now, the amendment remains part of the 2026 NDAA, and no further action has been taken by congressional committees. The situation remains a point of reference for lawmakers considering how to regulate AI assistance in the legislative process.
