Congress Grapples with AI Tool Restrictions Amid Public Demand

An internal memo from the Senate sergeant-at-arms, obtained by NPR, shows that Senate staff may use only three basic AI chat interfaces—Microsoft Copilot Chat, Google Gemini Chat for Workspace Enterprise Plus, and OpenAI ChatGPT Enterprise—at no cost to their offices. More capable “agentic” tools such as OpenAI’s Codex or Anthropic’s Claude Code, which can autonomously execute tasks and access files, remain barred. The sergeant-at-arms’ policy explicitly forbids platforms from independently accessing Senate drives, shared folders, email, Teams chats, or other internal resources.

NPR

Polls indicate a majority of Americans favor congressional regulation of AI. Industry leaders have warned of safety risks, prompting the current debate. Adam Kovacevich, founder and CEO of the tech-policy group Chamber of Progress, argues that lawmakers are drafting rules for technologies many have never used, creating a “problem” for effective policy.

NPR

Daniel Schuman, executive director of the nonpartisan American Governance Institute, emphasized the risk of data exfiltration:

NPR

• “The use of tools and technologies that have the possibility of exfiltrating data elsewhere are a significant risk for the Senate and for the House and elsewhere,” — Daniel Schuman

NPR

Across the Capitol, the House Office of the Chief Administrative Officer oversees AI use, approving tools from Microsoft, OpenAI, Google and Anthropic for specific scenarios.

NPR

Sen. Bernie Moreno (R-OH) described his office’s informal AI “staffer” named Albert, noting its utility for research:

NPR

• “He works long hours, doesn't require a lot of pay, but it's great for research, things that would normally take somebody a long period of time to kind of vet,” — Sen. Bernie Moreno

NPR

Moreno also highlighted the Senate’s prohibitions:

NPR

• “Right now, the Senate prohibits us from integrating outside systems into the Senate's,” — Sen. Bernie Moreno

NPR

Kovacevich criticized the cautious stance, suggesting that security concerns often serve as a “pretext” that slows progress and that private firms already manage sensitive data with advanced AI tools.

NPR

The restrictive environment may limit lawmakers’ firsthand experience with cutting-edge AI, potentially shaping legislation that leans heavily on precautionary measures. While some offices experiment with ad-hoc solutions, broader adoption hinges on whether Senate leadership relaxes the current “brakes” and authorizes more powerful models in the coming months.

NPR