OpenAI accounted for roughly $9 of every $10 spent on identifiable AI tools by US House offices during the year ending March 31, 2026, according to a CNBC analysis of House disbursement records. ChatGPT cost congressional offices $100,580 across 798 transactions. Anthropic’s Claude — the closest competitor — managed $13,160 across 37 transactions.
🔍 THE BOTTOM LINE: The same lawmakers drafting AI regulation have effectively chosen a default vendor before writing a single rule. OpenAI’s 88% dollar share and 96% transaction share means Congress is not just adopting AI — it’s standardising on one company’s product while that company lobbies for favourable treatment.
What the Numbers Show
CNBC identified at least $113,740 in spending on named AI vendors between April 1, 2025, and March 31, 2026. The figure excludes free accounts, AI bundled into broader software contracts, and most Senate usage — meaning the actual AI footprint on Capitol Hill is larger than these numbers suggest.
At least a sixth of House member offices appear in the spending records. TechCrunch reported that staffers use ChatGPT to draft memos, summarise legislation, prepare hearing materials, respond to constituents, and even draft social media posts.
Democratic offices accounted for $54,165 in identified AI purchases — more than three times the $15,782 spent by Republican offices. ChatGPT appeared in 44 Democratic member offices and 27 Republican offices. Claude appeared in just five Democratic offices and one Republican office.
The Influence Problem
The spending pattern creates what transparency advocates describe as a vendor lock-in dynamic. When congressional staff rely on one company’s tool to summarise legislation and prepare briefing materials, that company shapes how information enters the policy pipeline — not by lobbying, but by being the interface.
Rep. Julie Fedorchak, R-N.D., told CNBC that her staff used ChatGPT to build a searchable database of legislation and constituent materials, describing it as “saving dozens and dozens of hours every week of staff time.” That efficiency is real. But it also means the tool’s strengths, limitations, and presentation choices become embedded in how a member of Congress understands a bill.
The spending data lands at an awkward moment for AI policy. OpenAI is simultaneously racing for influence across Washington as lawmakers and federal agencies decide how AI should be purchased, used, and regulated. The company recently met with lawmakers on Capitol Hill in June 2026, with CEO Sam Altman speaking to reporters after meetings with members of Congress — the same month Amazon completed a $50 billion investment in OpenAI.
What Staffers Actually Do With It
Marci Harris, CEO of POPVOX Foundation, a nonprofit that trains congressional offices on AI use, told CNBC that staffers are using AI to summarise reports and bills, draft social media posts, prepare constituent responses, analyse legislation, and sort through policy research. She noted that AI is also “exacerbating” the staffer overload problem — lobbyists and constituents are using AI to send longer, more frequent, and more detailed material to congressional offices.
“Staffers are famously overwhelmed, and actually AI is exacerbating that,” Harris said. The irony: the same tool that saves time summarising legislation also means more legislation arrives faster, generated by AI on the other side.
Why Only Two Vendors
Google and Microsoft appear nowhere in the spending records despite both having AI products. The likely reason: Google’s Gemini and Microsoft’s Copilot are often bundled into existing enterprise contracts — Google Workspace and Microsoft 365 — and would not show up as separate line items in House disbursement records. The data captures only direct, identifiable AI purchases.
That means the 88% OpenAI figure likely overstates its true dominance if bundled AI is counted. But it also means the number of congressional offices using AI is higher than the spending records show. The records capture intent to purchase a standalone AI tool — which points to ChatGPT as the default conscious choice.
NZ Angle
New Zealand faces a similar question, though at smaller scale. When government agencies adopt AI tools for drafting, research, and constituent communication, they create the same vendor-dependency dynamic. The EU AI Act’s transparency rules that took effect August 2 now require disclosure when people interact with AI — but there is no equivalent obligation for governments to disclose which AI tools they use internally, or how much they spend. The US Congress data exists only because House disbursement records are public. Whether NZ’s Parliament publishes equivalent spending breakdowns is an open question worth asking.
❓ FAQ
Why does Congress spending $100K on ChatGPT matter? Because the same lawmakers writing AI regulation have effectively standardised on one vendor’s product. That creates a structural bias — the tool shapes how information reaches the lawmaker, not just how the lawmaker writes back.
Does this include free ChatGPT accounts? No. The CNBC analysis excluded free accounts, AI bundled into software contracts, and most Senate usage. The real AI footprint on Capitol Hill is larger.
Why did Democrats outspend Republicans 3-to-1 on AI? The data doesn’t explain why. Democratic offices may have larger staffs, more policy research needs, or a cultural willingness to adopt new tools. Some Democrats have also been more vocal about AI regulation, which may mean more engagement with the tools they’re trying to regulate.
Could this change? Yes. The data covers April 2025 to March 2026 — the first year of visible AI spending. As Claude, Gemini, and Copilot compete more aggressively for government contracts, the distribution could shift. The $50 billion Amazon-OpenAI deal and OpenAI’s multi-agent Washington push suggest the market for government AI is just starting.
🔍 THE BOTTOM LINE: Congress has a ChatGPT problem, and it’s not that staffers use it. It’s that the spending data reveals a structural dependency on one company before the rules are written. When the tool that summarises the legislation is made by the company lobbying for favourable legislation, the line between user and subject blurs.
