AI Is Quietly Becoming Part Of Congress’ Workforce While Lawmakers Are Still Figuring Out The Rules

A congressional staffer's accidental inclusion of an Anthropic Claude response in public legislative material has drawn attention to the growing use of artificial intelligence by lawmakers and their employees for official work.

The AI-generated text appeared in a summary accompanying an amendment from Rep. Anna Paulina Luna, R-Fla., to the National Defense Authorization Act. The document included the words "Claude responded," along with garbled material apparently copied from an interaction with Anthropic's chatbot.

Luna said her staff used Claude to review the summary rather than draft the underlying legislation. She went on to defend the use of the technology and said AI tools were already widely used by congressional staff.

The incident comes as generative AI is being adopted for a growing range of work on Capitol Hill, including policy research, legislative summaries, speeches and amendments.

The Congressional Research Service has identified potential congressional uses for generative AI including summarizing documents, drafting bills and speeches and generating other content. The agency has also warned that the technology can produce inaccurate information and amplify biases.

The House has moved to provide staff with officially approved AI tools. It purchased 6,000 Microsoft Copilot licenses for member, committee, leadership and institutional offices, according to a House Administration Committee report.

Copilot is integrated with Microsoft products already used across the House, including Outlook, Teams, Word and Excel. The committee said the system operates as an authorized cloud service and can handle sensitive House data, allowing employees to use AI without putting protected congressional information into public-facing chatbots.

The House is also developing AI applications specifically for congressional work. In July, the House Administration Committee launched an "AI Innovators Pipeline," which provides staff with a secure environment for developing custom applications with support from the House AI Center.

Projects developed through the program include tools for managing schedules, summarizing legislation and monitoring activity on the House floor, the committee said.

Separately, House officials are seeking funding for "Mia," or My Intelligent Assistant, a customizable AI agent intended to connect with House systems and data. The House Administration Committee said the system could help staff search for information and perform routine tasks across congressional platforms.

The adoption mirrors a broader expansion of AI across the federal government. The Government Accountability Office found that reported generative AI use cases at 11 selected federal agencies increased from 32 in 2023 to 282 in 2024. Overall reported AI uses at those agencies nearly doubled from 571 to 1,110 during the same period.

Congress' adoption of AI is accelerating as lawmakers simultaneously consider how the technology should be regulated.

House Democrats this week sought information from OpenAI and Anthropic following reports that AI agents escaped containment during security testing and accessed external systems. The lawmakers raised questions about safeguards surrounding increasingly capable models and possible national security implications, Reuters reported.

President Donald Trump has meanwhile opposed tighter federal restrictions on the industry, warning that lawmakers could regulate AI companies "out of business." Congress has yet to adopt comprehensive federal legislation governing artificial intelligence, while states have moved ahead with their own rules in areas including transparency and discrimination.

The House established its own chamber-wide AI policy in 2024. The framework covers the use of AI systems by House offices and was designed to be updated as the technology and its risks evolve.