ChatGPT Takes Early Lead In Congress AI Spending
CNBC found ChatGPT captured about 88% of identifiable House AI-vendor spending in the year to 31 March 2026, even as Congress weighs AI purchasing and regulation.
A narrow slice of congressional AI spending already points to a strong default supplier before federal lawmakers finish writing rules for the technology.
The Next Web reported on 4 August 2026, citing CNBC's review of House disbursement records, that ChatGPT received nearly nine of every ten visible dollars from the $113,740 spent on named AI vendors in the year that ended in March.
According to the CNBC figures, OpenAI's tool accounted for about 88% of that vendor spending and 96% of transactions, leaving Anthropic's Claude far behind.
That gap suggests congressional offices are standardising staff habits around one product while the legislative debate over AI procurement, safety and disclosure remains unsettled.
Spending Records Show A Workflow Default
ChatGPT appeared in at least 71 House member offices, roughly one in six, according to the CNBC analysis.
The records described ordinary staff functions rather than frontier research: bill summaries, memo drafting, hearing preparation and constituent responses.
Representative Julie Fedorchak, a North Dakota Republican, described a ChatGPT-built database of her legislative work as saving staff “dozens and dozens of hours every week.”
That workflow evidence is commercially important because public-sector AI adoption often begins through routine productivity tools.
Once staff templates, training materials and internal examples are built around one model, a later switch requires more than a new software purchase.
Party Spending Adds A Policy Tension
The disbursement records also show a political split that complicates the public debate over AI risk.
According to the CNBC analysis , Democratic offices spent $54,165 on identifiable AI tools, compared with $15,782 by Republican offices, and ChatGPT appeared in 44 Democratic offices and 27 Republican offices.
Democratic lawmakers have also been among the more visible voices warning about AI's effects on workers, privacy and elections.
The source record does not show that those offices are using AI in sensitive casework or policy drafting beyond the categories described in the analysis.
Skepticism about the industry has not prevented paid adoption inside congressional operations, creating a lobbying opening for vendors before formal procurement standards arrive.
Vendor Access Becomes Part Of Governance
OpenAI's lead was not only a product result. the House distributed its first 40 ChatGPT licences in 2023 before restrictions arrived in 2024, and cited Marci Harris of the POPVOX Foundation saying ChatGPT had become almost synonymous with AI for staff.
According to TNW, OpenAI put $1m into first-quarter Washington lobbying, an 82.5% year-on-year increase.
Claude's weaker showing carries a different governance signal. interest in rival tools rose after the Pentagon identified Anthropic as a supply-chain risk in March, while the Senate bars Claude for official use.
Those restrictions mean spending records can reflect access policy as much as user preference.
The figures remain incomplete because the records miss free accounts, AI bundled into broader software such as Microsoft Copilot, reimbursements and most Senate activity.
Even so, the visible House pattern shows why procurement habits matter: the institution preparing AI rules is already learning the technology through a supplier whose product and lobbying are embedded in daily office work.
