In detail
- A screenshot showed Claude prompts in a summary for the 2027 National Defense Authorization Act amendment.
- Luna corrected her statement multiple times: first saying staff used AI to 'correct a draft text,' then clarifying that only the summary, not the bill text itself, was AI-checked.
- She emphasizes that 'NO Legislation is ever drafted with AI' and that all bill text comes from the House Legislative Council, which is prohibited from using AI.
- The screenshot is an AI-generated summary artifact, not the actual bill text.
Why it matters
This highlights the growing gray zone around AI use in regulated contexts: even in US lawmaking, confusion arises between legitimate helper functions (spellcheck) and prohibited text generation. For businesses, this is a warning to clearly document AI use in compliance-critical processes.
For you Establish clear guidelines for where AI can and cannot be used in your organization — especially in legal, regulatory, or security-critical documents.