Senators can talk about artificial intelligence all day, but their own staff can’t use some of the more advanced AI tools. That’s the situation on Capitol Hill, where Senate lawmakers are shut off from advanced AI software even as the public demands Congress regulate the technology.
An internal communication sent to Senate staff in the spring by the sergeant at arms, who oversees cybersecurity in the chamber, bars staff from using anything beyond the most basic AI interfaces. Senate staff have access to three free AI chat interfaces: Microsoft Copilot Chat, Gemini Chat for Google Workspace Enterprise Plus and OpenAI ChatGPT Enterprise. More advanced tools, like OpenAI’s Codex or Anthropic’s Claude Code and Cowork, are not authorized.
Spring Message From The Sergeant At Arms
The message from the sergeant at arms, sent in the spring, laid out the rules. Staff are barred from using all but the most basic AI tools. The approved interfaces are provided at no cost to their offices.
The restriction applies to tools that allow “agents” powered by advanced AI models to complete tasks autonomously at a user’s request. Platforms cannot independently access internal Senate drives, shared folders, email, Teams chats or other Senate resources.
Daniel Schuman, executive director of the American Governance Institute, says the risk of data leaving Senate systems is significant. He also says the incentives of the institutional setup slow progress.
Bernie Moreno’s Office Uses AI
Sen. Bernie Moreno, Republican of Ohio, says his office uses AI and has an AI staffer named Albert. Moreno says the Senate prohibits integrating outside systems into its own. He hopes the prohibition gets lifted next year.
The senator’s office is one known example of AI use within the Senate. The source does not describe how other offices handle AI.
The Role Of The Sergeant At Arms
The sergeant at arms is responsible for cybersecurity in the chamber. That role carries weight when it comes to AI approval.
Adam Kovacevich, founder and CEO of Chamber of Progress, says Congress has always lagged behind. He adds that security arguments are often pretextual objections.
Schuman points to the sergeant at arms as the person most likely to hold back on AI approvals. “The sergeant at arms is going to be more reticent because they don’t want to have the finger pointed at them in case things go wrong,” he says.
House Rules On AI
Across the Capitol, the House AI use is overseen by the Office of the Chief Administrative Officer. House rules outline specific use cases where AI is permissible.
Microsoft, OpenAI, Google and Anthropic tools are approved in various House scenarios. The CAO office did not immediately reply to a request for comment.
The Senate Committee on Rules and Administration, which oversees the sergeant at arms, did not address specifics in response to NPR questions. A committee spokesperson said policies are designed to protect Senate data with stronger security and contractual requirements.
Public Demand For Regulation
A majority of American public wants Congress to regulate AI. That demand exists alongside the Senate’s own restrictions.
The contradiction is clear. Lawmakers can debate AI regulation, but their own offices are operating under rules that prevent staff from using the very tools they are considering.
Data Exfiltration Concerns
Schuman raises a specific concern about the approved interfaces. He says the risk of data leaving Senate systems is significant.
The approved interfaces cannot independently access internal Senate drives, shared folders, email, Teams chats or other Senate resources. That is the stated protection. Whether it is enough to satisfy the sergeant at arms remains an open question.
The committee spokesperson’s statement about protecting Senate data with stronger security and contractual requirements does not address the specific tools that are off-limits.
The Scope Of The Ban
The Senate’s position puts it behind the curve. Kovacevich’s assessment that Congress has always lagged behind holds for AI as well as for other technology.
The restriction is not about banning AI entirely. It is about limiting which tools staff can use. The approved interfaces are provided at no cost to offices.
But the advanced tools are not available. That means Senate staff cannot use OpenAI’s Codex for coding, Anthropic’s Claude Code for complex reasoning or Anthropic’s Cowork for task automation.
Moreno’s Hope For Change
Moreno’s hope that the prohibition gets lifted next year suggests the restriction is temporary. His office already uses AI, and he has an AI staffer named Albert.
The senator’s office operates under different rules than the general description of Senate staff. The source does not describe how other offices handle AI.
| Office | AI Oversight | Approved Tools |
|---|---|---|
| Senate | Sergeant at arms | Copilot Chat, Gemini Chat, ChatGPT Enterprise |
| House | CAO | Microsoft, OpenAI, Google, Anthropic |
Where the paper stands
The paper backs narrow rules against direct harm, such as forcing companies to disclose safety failures they hid, and is against broad rules that hand the market to the incumbents. It is also against letting big companies hide their failures. The Senate’s own ban on staff using advanced AI tools, while understandable in terms of data security, locks out the very staff who could help lawmakers understand the technology they are regulating. The paper would prefer a lighter, targeted approach — disclosure requirements, safety reporting, and narrowly tailored guardrails — rather than a blanket prohibition that freezes everyone in place.
The irony here is that the Senate, which is supposed to represent the public’s desire for regulation, has made its own staff unable to use the tools they need to evaluate AI’s capabilities. That is a problem that needs fixing.
The paper would want to see the Senate Committee on Rules and Administration loosen its grip on AI approvals, allowing staff to use more capable tools while keeping real protections in place. Lawmakers should be able to test the technology themselves before they vote on it.
Key Facts Box
- Senate staff have access to three free AI chat interfaces: Microsoft Copilot Chat, Gemini Chat for Google Workspace Enterprise Plus and OpenAI ChatGPT Enterprise
- Advanced tools banned: OpenAI’s Codex, Anthropic’s Claude Code, Anthropic’s Cowork
- Moreno hopes the prohibition gets lifted next year
- A majority of American public wants Congress to regulate AI
The Senate’s own approach to AI is not described consistently in the source. The senator’s office operates under different rules than the general description of Senate staff.
The public wants Congress to regulate AI. Lawmakers can talk about it all day. Their staff can’t use the tools that might help them do the work.
Source material: “As Congress confronts AI, Senate lawmakers are shut off from its most advanced tools,” NPR.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.