What’s happening with AI and why should I care?
POPVOX Foundation’s “Future-Proofing Congress” is a newsletter created for the busy Congressional staffer to keep you updated (and inspired!) about operational improvements and new resources on and off the Hill. Learn more and subscribe here.
Helping you stay up to date and ahead of the curve
Spring is in the air, shamrocks are around the corner, and we’re rolling into yet another season with even more technology news that can feel difficult to understand and keep up with. Trust me, I get it. I saw this 1977 Honeywell ad a few weeks ago, and it really summed up how I feel about this constantly changing part of our world.
But that’s what Future-Proofing Congress is for – to help you stay in the loop, share some new ideas to implement for your office, and make sure you’re up-to-date to help you do your job even better. That’s why POPVOX Foundation has been sitting down with House and Senate offices for real, off-the-record discussions about AI and how staff can be using these tools to level up their workflows and productivity. While Congress can be a place of modernization and innovation (yes, really), it is also a place of tradition and rule-following. This means Members and staff do things “the way they’ve always been done,” which can leave little room for new tools and technology – even if those tools and technologies have broken through to the rest of the world.
In this month’s edition, we will:
Review some recent AI developments to help you stay on top of the news cycle
Share information about our team’s AI training for Congressional offices (TL;DR: they are free, customizable, and not sponsored by any tech companies)
Highlight some recent Congressional modernization wins
Share a proposal our team has worked on for institutionalizing AI resources in a way that boosts Congressional capacity
Highlight a new newsletter from our team exploring the cultural and technical gap between DC and Silicon Valley
Take it from me, you don’t need to be a tech person to care about this! You just need to know enough to ask the right questions and be open to learning.
Have a great week,
Danielle Stewart
Advisor for Congressional Initiatives
POPVOX Foundation
What’s Happening: The Short Version
We know you are busy. Here’s a short and sweet recap of what you actually need to know about what is happening in the world of AI.
The Senate Just Approved Three AI Tools for Official Use
This is the big one. On March 9, the Senate Sergeant at Arms’ Chief Information Officer announced that three AI tools are now authorized for use with official Senate data:
Microsoft Copilot Chat
Google Workspace with Gemini Chat
OpenAI’s ChatGPT Enterprise
These are the first tools to receive “Tier 2” authorization under the Senate’s AI governance framework, meaning Senate staff can use them with official Senate data — not just for general research. The New York Times reported the news on March 10, citing the internal memo.
Our team at POPVOX Foundation published an updated guide to both the House and Senate AI use policies — including what’s approved, what the frameworks look like, and what staff should know. The NYT story cited POPVOX Foundation’s review of both chambers’ guidance. If you haven’t read it yet, it’s worth a look.
A note on what’s not on the Senate list
You may notice that Anthropic’s Claude is not among the Senate’s approved tools. Claude is approved for use in the House. The Senate’s decision reflects its own security review and procurement process — the Senate’s Tier 2 approvals cover tools that have met specific contractual and cybersecurity requirements through the SAA’s review process. While Anthropic is in the news right now for a separate dispute with the Department of Defense over the use of its AI tools for military purposes, that controversy is a distinct matter from the Senate’s internal technology decisions. The Senate’s approved tool list is the product of its own independent evaluation process.
AI Alphabet Soup
Even in a city full of acronyms and code names, it can still be hard to keep track of the major AI companies, the models available for use, and which ones are approved for use in the House and Senate. Each tool offers a different “personality” and can process information in different ways.
If you have questions on which tools and use cases are approved for use in Congress, you’re not alone. POPVOX Foundation Director of Government Innovation and Global Initiatives, Aubrey Wilson, published a helpful guide here on Congressionally-approved tools, use cases, and policies.
GPT-5.4 (OpenAI/ChatGPT): The newest and most capable model from OpenAI, released March 5. It’s designed for professional work — research, document drafting, coding, and complex reasoning — and is the first OpenAI model with the ability to autonomously operate computers and navigate software. OpenAI says it hallucinates less than prior versions. ChatGPT is approved for use in both the House and the Senate.
Gemini 3.1 Pro (Google): Leads on multimodal tasks (processing text, images, audio, and video together). Largest context window (can ingest roughly 750,000 words at once, useful for large document review). Google’s free NotebookLM tool, which can take any uploaded document, including bill text, and generate a podcast, video, or slide deck from it, now runs on Gemini 3.1 Pro. Gemini is approved for use in the House and the Senate through the Google Workspace.
Claude 4.6 (Anthropic): Strong at coding, document analysis, and sustained autonomous work. Emphasizes safety. Three tiers: Opus (most capable), Sonnet (best value), Haiku (fastest/cheapest). Claude is approved for use in the House, but is not currently approved for use in the Senate.
Microsoft Copilot: Now rolling out across the House with 6,000 enterprise licenses, and newly approved in the Senate. Integrated into Microsoft 365 tools you already use — Word, Excel, Outlook, Teams. Copilot is approved in both chambers.
Grok 4.1 (xAI): Strong for reasoning, real-time information via X/Twitter integration. Grok is not approved for House or Senate use.
Need to Know - Agentic AI
A major shift is underway. For a while, when you thought about using AI, it meant talking to a chatbot (think: ChatGPT or a customer service bot online). Now, AI tools are searching the web, filling out forms, making calls, and completing multi-step tasks autonomously. Companies call these “agentic” AI systems. These systems run and execute tasks without a human in the loop. The AI plans, acts, observes, and adjusts until the task is complete.
Generative AI uses a prompt to create a product, answer a question, or provide deep research for the user. Agentic AI can manage full workflows, solve problems, and take action.
TL;DR: these models can do SO much more than just answer questions in a chatbot. Learn how to enhance your workflow with our job-specific resources here.
AI on the Campaign Trail - Already Here.
AI isn’t a future election concern — it’s showing up in races right now. In Chicago’s 2nd Congressional District primary, a campaign ad for Jesse Jackson Jr. features an endorsement from former Rep. Bobby Rush whose vocal cords were damaged by throat cancer. The ad uses AI to restore Rush’s voice to deliver the endorsement. Rush consented and has used the technology for podcasting and interviews. But the ad has sparked debate: When does AI voice restoration cross from accessibility tool to campaign manipulation? Meanwhile, an AI-backed super PAC has entered the race supporting Jackson, while a rival super PAC opposing him advocates for tougher AI regulation — turning AI governance itself into a campaign issue.
AI-generated deepfakes (realistic fake audio and video) of candidates are appearing in campaign materials across the country. States are moving quickly: Pennsylvania, Maine, and others are advancing disclosure requirements for AI-generated campaign content, and YouTube just announced it is expanding its deepfake detection tools to cover government officials, political candidates, and journalists. But there is still no comprehensive federal legislation on AI-generated content in elections, and the FEC has not established rules specific to AI-generated campaign materials.
Separately, an analysis circulating on social media ran Congressional press releases through AI/plagiarism detection tools and drew conclusions about which offices are using AI to draft statements. A word of caution: current AI detection tools are unreliable and produce significant false positives. Formal, policy-oriented writing often triggers these detectors even when no AI was involved. Take these analyses with a large grain of salt.
This serves as a good reminder that as this technology and its widespread usage change, so must our policies and awareness of these tools. We have resources for safety dos and don’ts here.
Quick Hits: What Else You Should Know
Meta acquired Moltbook, the viral social network built for AI agents — not humans. On Moltbook, AI bots post, comment, and interact with each other autonomously.
AI may create a “demand machine” for government services. A new report from New America warns that instead of reducing workloads, AI tools that make it easier for constituents to interact with government could actually increase demand for services — surfacing needs that were previously hidden by friction and bureaucratic complexity. Worth thinking about for casework operations.
A music executive built an AI situational awareness tool that reportedly rivals capabilities of intelligence agencies — a reminder that powerful AI tools are no longer confined to governments and large institutions. The barriers to entry for sophisticated analysis are dropping fast.
The Cato Institute argues we need to “put AI to work for the American people” — a piece by Kevin Frazier calling on young Americans to seize AI’s economic opportunity, including through the new US Tech Corps initiative.
Brazilian diplomat Tiago R Santos built “Hatsell” (hatsell.app) to help run meetings according to Robert’s Rules of Order
AI Training Tailored to Your Office
Many Congressional staff are already using institution-approved AI tools to draft social media content, research policy, analyze legislation, and improve workflows. AI is an evolving technology with changing capabilities, and no one has all of the answers to how these tools can be used and in what contexts yet.
POPVOX Foundation offers free, off-the-record, personalized AI training designed specifically for Congressional offices. We have helped dozens of offices become more familiar with these tools and open discussions about how and where they can be used to streamline existing workflows.
What makes our training different:
Taught by former Congressional staff (DC and District) who understand actual workflows and use cases.
Your pace, your format — office-wide sessions, virtual or in-person, for DC and district staff. These are direct team meetings where individuals feel comfortable asking real questions and applying learning to their specific office’s AI use policy.
Completely nonpartisan and not tech-company backed — POPVOX Foundation does not receive funding from the technology industry. Our trainings are transparent and realistic about both usage potential and risks.
Aligned with House and Senate guidance on AI usage.
Customized to your needs — from AI 101 to advanced applications, tailored by role (communications, leg shop, casework, and more).
Interested? Reply to this email or reach out directly to Aubrey Wilson, Director of Government Innovation & Global Initiatives: Aubrey@popvox.org
Browse POPVOX Foundation’s AI guides and resources for Congressional staff at popvox.org/ai.
Modernization Wins: By the Numbers
You may not know it from the media, but there is good news about Congress: the programs and tools that have emerged from years of modernization work are gaining real traction. Here’s where things stand:
Legidex
Launched last June to fulfill a Select Committee on Modernization recommendation, Legidex is a real-time staff directory built on payroll and Office of the Clerk data. It now includes House, CRS, and GAO staff, with Senate conversations ongoing. Since launch: 1,000 users per day, 6,000 users per month, 20,000 sessions per month — and growing.
Microsoft Copilot
Over the summer, the CAO and Modernization Subcommittee procured one-year Copilot licenses at no new expense to the government, paired with House-specific role-based training and prompt guides. The CAO is running more than 5 virtual training sessions per week. A recent all-House staff briefing drew over 150 attendees — standing room only. As of this writing: more than 180 Member offices have opted in, a dozen committees are in the process of allocating licenses, and all House institutional offices are enrolled. Offices receiving licenses are also required to update their AI policies in their handbooks.
Note: Some offices have reported that Copilot’s content moderation defaults sometimes refuse requests that are political in nature — an early illustration of a mismatch between commercial AI tools built for general enterprise use and the inherently political work of Congress. This is something to watch as more offices come online.
Retention through Educational Advancement Program (REAP)
Expanded in January 2025 after 18 months of work, REAP now covers private student loans, professional development expenses, and credentialing costs, not just federal student loan repayment. By the end of 2025, more than 2,400 staff participated, including 366 who used the expanded options.
Proofreading Waiver Pilot
Since September, Members and Committees can opt to waive GPO proofreading when introducing new legislation, reducing processing time and allowing bills to post on Congress.gov faster. Of 1,251 bills introduced since the pilot launched, 219 waived proofreading (about 17.5%) — and the majority of those were posted on Congress.gov the same day.
2,800 Miles
POPVOX Foundation’s Program Associate Caitlin McNally, just launched a newsletter called 2,800 Miles — named for the distance between San Francisco and Washington, DC. The newsletter explores the cultural and professional gap between Silicon Valley and the policy world, written by a former Hill staffer now living in San Francisco. If you’ve ever tried to explain your job to a tech person, this one’s for you. Subscribe here.
What Comes Next: A Proposal to Make This Permanent
The modernization wins happened because people inside and outside Congress pushed for dedicated resources, smart investments, and institutional follow-through. But as AI capabilities continue to accelerate, the question is whether the institution has the internal support structure to help Members and staff adopt these tools confidently, responsibly, and at the pace the moment demands.
That’s the idea behind a new proposal from POPVOX Foundation: the Congressional Office of Capacity and Technology (C-TECH). C-TECH would be an independent, nonpartisan office within the House — modeled on the Office of the Whistleblower Ombuds — dedicated to helping Congressional staff actually use AI and emerging technology well. Not IT infrastructure. Not policy research. Think: training, resources, change management, and a judgment-free place to ask questions and get support.
The proposal outlines a phased approach, starting with a lean six-person team focused on adoption support and trust-building in the House, and scaling over time to serve both chambers with intelligence monitoring, embedded committee experts, and a Congressional Chief AI Officer.
Why does this matter for you? If you’ve ever felt like you’re expected to figure out AI on your own, or wished there were someone in-house whose job it was to help you integrate these tools into your actual workflow — that’s exactly what C-TECH is designed to be.
POPVOX Foundation has published the full proposal, including staffing, budget, and appropriations language for the FY2027 Legislative Branch Appropriations cycle.
Resources Made for You, from POPVOX Foundation
Supporting Constituents in Crises Overseas
For offices new to casework or new to handling overseas crisis situations, POPVOX Foundation’s Casework Navigator program has a quick guide to how offices can respond to requests for urgent help in an overseas crisis.
Senate Pro Forma Tool
POPVOX Foundation highlighted the Senate’s Pro Forma tool, a resource for Senate staff used by 12 committees that handles everything from meeting scheduling and amendment tracking to nomination processing and classified document management – all in one integrated system.
AI Resources for MPs and Parliamentary Staff
POPVOX Foundation has compiled AI guides and resources specifically designed for legislative staff working in parliamentary settings around the world. This can provide helpful context for anyone tracking how other legislatures are approaching the same questions we are grappling with here at home.
How Brazil turns citizen ideas into bills
POPVOX Foundation Fellow Beatriz Rey writes in ModParl about the Brazilian Senate’s new AI tool that plugs the public’s submitted ideas directly into the workflow of legislative consultants — including ideas that never reached the signature threshold to be formally considered. A quiet but significant example of what expanded constituent input could look like.
Worth Reading
How a Congressional Office Actually Works
Statecraft publishes in-depth interviews with current and former government officials on how their institutions actually function day-to-day. This piece is a useful read for new staff getting oriented and for experienced staff who want a clear-eyed outside perspective on how offices are structured, what drives them, and where the common friction points are.
A Day in the Life of the House Floor
Congressional scholar Matt Glassman walks through what actually happens on the House floor on a typical legislative day, like the procedural choreography, who’s in the room and why, and what the public (and most constituents) never see. Useful background reading for staff who want to understand how floor action maps to the work happening in their office.
About POPVOX Foundation
POPVOX Foundation is a nonpartisan nonprofit that helps democratic institutions keep pace with a rapidly changing world. Through publications, events, prototypes and technical assistance, the organization helps public servants and elected officials better serve their constituents and make better policy.