US Congress staff accidentally paste AI-generated text into defense bill

US Congress staff made a mistake by copying a confusing text generated by Claude into the public record of the National Defense Authorization Act. This incident, reported by The Washington Post, shows that AI tools are widely used on Capitol Hill but lack proper oversight.

Representative Anna Paulina Luna's office from Florida admitted the error. Luna said many staff use ChatGPT, Claude, and Grok, and she personally prefers Grok. She doesn't think using AI for summaries is against the law.

Currently, there are almost no specific rules for AI use in Congress. Existing rules are scattered and loose, and no staff member has been disciplined for using AI. Some assistants set their own limits, like Sydney Bloom, an aide to Representative Cleo Fields from Louisiana, who decided never to write documents from scratch with AI. But such self-regulation isn't common in Congress.