Senate AI Policy Blocks Staff From Advanced Tools Lawmakers Aim to Regulate

Senate lawmakers tasked with regulating artificial intelligence are operating under internal cybersecurity policies that prevent them and their staff from using the most advanced AI systems currently under debate, according to internal communications obtained by NPR.

The Senate Sergeant at Arms has authorized three AI chat interfaces for staff use at no cost: Microsoft Copilot Chat, Gemini Chat for Google Workspace Enterprise Plus, and OpenAI ChatGPT Enterprise. However, more capable autonomous agent tools—including OpenAI's Codex and Anthropic's Claude Code and Cowork—remain unauthorized despite being central to regulatory discussions in Washington.

These advanced systems, which can complete tasks autonomously and access computer files with varying permission levels, represent a different category of capability than the web-based chat interfaces currently available. Senate rules explicitly state that approved platforms "cannot independently access internal Senate drives, shared folders, email, Teams chats, or other Senate resources."

Why it matters

The restriction creates a knowledge gap at a critical moment. A majority of Americans want Congress to regulate AI, yet policymakers are writing rules for technology many have never used. This dynamic risks producing legislation that focuses exclusively on limiting harms while missing potential benefits—what one industry advocate described as "all brakes and no steering wheel." The gap is particularly significant as AI industry leaders themselves have raised safety concerns about the very tools Congress cannot access.

Security concerns versus practical experience

Daniel Schuman, executive director of the American Governance Institute, acknowledged that tools capable of data exfiltration pose legitimate risks for the Senate. Yet he noted that similar concerns apply to widely used technologies like Google Docs and email.

"Those concerns are not the best reason for the non-approval of certain technologies," Schuman said. "I think people want to do better, but the incentives of the way the institution is set up is a drag on progress."

Adam Kovacevich, founder and CEO of Chamber of Progress, a center-left tech policy group, argued that security objections often serve as pretexts for institutional inertia. "You've got Fortune 500 firms who have plenty of sensitive customer information and data using all these tools and using them successfully," he said.

A Senate staffer speaking anonymously confirmed that advanced AI tools are being vetted for specific use cases but offered no timeline for approval.

House takes different approach

The House of Representatives operates under separate rules managed by the Office of the Chief Administrative Officer. According to a House staffer, that chamber has approved tools from Microsoft, OpenAI, Google, and Anthropic for various specific scenarios, though details of those use cases were not provided.

Offices finding workarounds

Some Senate offices appear to be pushing boundaries. Senator Bernie Moreno, Republican of Ohio, told technology news show MTS Live that his office employs what he calls an AI staffer named Albert for research tasks. Moreno expressed frustration with Senate restrictions on integrating outside systems and said he hopes rules will change next year. His office did not respond to questions about whether Albert complies with current Senate policies.

The Senate Committee on Rules and Administration, which oversees the Sergeant at Arms, defended the policies as necessary to "protect Senate data" with "stronger security and contractual requirements," while acknowledging that offices should have AI access "subject to adequate safety and security checks."

These details were first reported by NPR.