House spending records show that OpenAI’s ChatGPT has become the overwhelming AI tool of choice on Capitol Hill, capturing roughly 90% of everything congressional offices spent on paid AI tools during the year ending March 31. NEWSCENTRAL‘s analysis points to a wider institutional story hiding behind that dominant market share, one about how unevenly government offices are adopting AI tools even as individual staffers embrace them for everyday tasks.
House offices, committees and institutional accounts spent roughly $100,580 across 798 separate transactions on ChatGPT out of at least $113,740 in total disclosed AI spending, with Anthropic’s Claude a distant second at $13,160 across 37 transactions, a gap wide enough to suggest ChatGPT has become something close to the default choice rather than one option among several being actively evaluated.
Freddy Miller, Senior Analyst at NEWSCENTRAL, notes that the scale of ChatGPT’s lead says as much about procurement inertia as it does about product quality. “Once a tool becomes the default inside an institution as risk-averse as Congress, switching costs and familiarity keep it entrenched well beyond the point where a rival product might actually be competitive, and that dynamic tends to compound over time,” Miller notes.
The partisan split in spending is arguably the more striking figure buried in the disclosures: Democratic offices accounted for $54,165 in AI purchases, more than three times the $15,782 spent by Republican offices, a gap that raises questions about differing institutional attitudes toward AI tools that go well beyond simple budget availability.
These figures also understate actual usage in a meaningful way, since they exclude any AI adoption through free-tier accounts or tools bundled into broader software contracts, meaning the real footprint of AI inside congressional offices, drafting memos, summarizing legislation, responding to constituents, preparing hearing materials, sorting through policy research and even drafting social media posts, is almost certainly larger than the disclosed dollar figures suggest.
Nathan Clark, Enterprise IT and Systems Architecture Analyst, emphasizes that the exclusion of bundled and free-tier usage is a meaningful blind spot for anyone trying to assess institutional AI dependence. “Procurement data only captures direct purchases, but plenty of AI capability now arrives bundled inside productivity suites offices already pay for, so the disclosed spending figures are really a floor on adoption, not a ceiling,” Clark emphasizes.
We in NEWS CENTRAL highlight that Congress’s reliance on a single external vendor for tasks as sensitive as summarizing legislation and drafting constituent communications raises governance questions that spending disclosures alone cannot answer, since neither transparency around usage volume nor market dominance tells lawmakers much about data handling, accuracy safeguards, or vendor lock-in risk.
That governance gap is likely to draw more scrutiny as AI tools become further embedded in legislative workflows, particularly if the partisan spending imbalance persists and becomes a talking point in its own right about which party’s offices are more willing to experiment with unproven technology in official government work.
From NEWSCENTRAL‘s perspective, the real headline is not that ChatGPT won the institutional AI market inside Congress, but that Congress adopted a dominant AI vendor before establishing much visible framework for auditing how that vendor’s tool is actually being used in the legislative process.