A chatbot-generated summary, complete with timestamp, got pasted directly into the public record of an NDAA amendment process by a staffer in Rep. Anna Paulina Luna’s office. Nobody stopped it. Nobody was disciplined. According to the Washington Post, almost nobody was surprised.
That context anchors where Congress actually stands on artificial intelligence.
The Tools Are Already Inside
AI tools including ChatGPT, Claude, and Grok are already handling congressional work — speeches, mail sorting, hearing prep, and amendment drafting — across both chambers.
The House didn’t stumble into this quietly. It purchased 6,000 Microsoft Copilot licenses and actively encouraged staff adoption. Luna told the Washington Post that tools like ChatGPT, Claude, and Grok are commonly used, and she didn’t view AI-generated summaries as illegal.
That’s not a scandal. That’s Tuesday on Capitol Hill.
The appeal is straightforward: aides describe AI as saving hours on research, communications, and legislative prep. When the legislative calendar moves fast, staffers use what works.
Here’s what internal rules technically prohibit, according to the Washington Post:
- Using sensitive constituent information in AI prompts
- Generating deepfakes
- Making personnel decisions via AI
- Finalizing legislation with AI output
Enforcement, however, exists only on paper. The Washington Post found no formal disciplinary case against any staffer for breaking these rules. People familiar with House operations were unaware of any enforcement action by the chief administrative officer. The rules exist. The consequences, reportedly, do not.
The Regulator Is Also the User
Congress is drafting consumer-protection bills targeting the exact chatbot tools its own staff uses daily — and the credibility gap keeps widening.
While Congress debates guardrails for everyone else, it’s simultaneously advancing the CHATBOT Act (S.4407) in the Senate and H.R. 7985 in the House — both aimed at chatbot-related consumer protections. Think of a referee throwing flags for holding while running the exact same play on the side. The conflict isn’t hidden. It’s just unaddressed.
The Washington Post frames the adoption as driven by convenience and competitive pressure — aides keeping pace with fast-moving legislative demands. That’s understandable.
What’s harder to defend is writing rules for an industry that an institution can’t govern within its own offices. The question worth asking isn’t whether government health, financial, and legal data guardrails drafted under these conditions can realistically be trusted — by anyone outside the Capitol, or inside it.
