AI is being widely used in the US Congress to draft speeches, press releases, prepare hearing questions, process constituent mail, and assist in drafting bills.
An aide to Representative Anna Paulina Luna accidentally copied an entire response from Claude, including timestamps, into official congressional documents.
Ms. Luna stated that staff use of ChatGPT, Claude, or Grok is very common and does not violate current laws or regulations.
Both the House and Senate have permitted the use of multiple chatbots such as Copilot, ChatGPT, Gemini, and in the House, Claude as well.
The House purchased 6,000 Microsoft Copilot licenses from the previous year and encourages staff to use them.
Current regulations only prohibit AI from processing sensitive constituent data, creating deepfakes, making personnel decisions, or finalizing bills.
A chatbot can assist in drafting bills or responding to constituents, but it must not be the entity performing the final step before enactment.
The Washington Post reported that almost no one is fully aware of these regulations, and no disciplinary cases for violations have been recorded.
Some congressional aides use AI to draft budget amendments, propose hearing witnesses, manage press relations, or analyze media data.
One aide even uploaded an entire contact list and years of message history into Claude to build a working database.
Some offices have begun training AI on lawmakers’ speech styles to generate content matching personal writing voices.
Senator Elissa Slotkin requires staff to use AI strictly as an aid, prohibits generating final speeches, mandates AI training participation, and accepts unannounced audits.
Some lawmakers warn that AI should assist humans rather than replace the responsibilities of elected representatives.
📌 AI has rapidly become an everyday tool in the US Congress, while regulatory frameworks remain rudimentary and enforcement is virtually negligible. AI currently assists with everything from policy research to bill and speech drafting, yet the boundary between human assistance and replacement remains undefined. This raises questions about accountability, transparency, and the legislative body’s reliance on AI systems developed by tech companies themselves.
