The data provides only a partial view of congressional AI adoption. It excludes free AI accounts, AI capabilities bundled into broader software subscriptions, reimbursements that do not identify specific software, and nearly all Senate usage because Senate spending reports do not disclose vendors in the same way.
Even with those limitations, the records suggest AI tools have become part of everyday congressional work. Staff are using them to summarize legislation, prepare briefing materials, draft constituent communications, analyze policy research, write social media posts, and assist with hearing preparation.
Rep. Julie Fedorchak, R-N.D., said ChatGPT has significantly reduced the amount of staff time required to prepare for meetings and legislative work. Her office created a searchable database of legislation, constituent communications, and other materials that AI can use to generate briefing memos.
Fedorchak said the system is "saving dozens and dozens of hours every week of staff time."
Marci Harris, CEO of the nonprofit POPVOX Foundation, said AI has become increasingly valuable for congressional staff facing growing workloads. At the same time, she noted that AI is also increasing the amount of material offices receive because lobbyists and constituents are using the technology to produce longer and more detailed submissions. "Staffers are famously overwhelmed, and actually AI is exacerbating that," Harris said.
The spending records also reveal a partisan difference in visible AI purchases. Democratic House offices accounted for $54,165 in identified AI spending, compared with $15,782 by Republican offices. ChatGPT subscriptions appeared in 44 Democratic offices and 27 Republican offices, while Claude purchases were recorded in five Democratic offices and one Republican office.
CNBC's review notes that these totals likely understate actual AI use. The figures do not include Microsoft's Copilot when it is bundled into enterprise software agreements, nor do they capture other AI features integrated into existing workplace software.
House officials also emphasize that their AI strategy is not tied to a single provider. Samantha Carter, chief communications officer for the House Chief Administrative Officer, said the chamber's approach "is deliberately not limited to a single vendor."
Congress began formally experimenting with generative AI in 2023, when the House Digital Service launched an AI working group and distributed its first 40 ChatGPT Plus licenses to bipartisan staff members. By the time the House introduced more restrictive guidance governing AI use in September 2024, ChatGPT had already become the best-known AI product among many congressional offices.
"OpenAI was the firm that they heard about," Harris said. "In the early days, ChatGPT was almost synonymous with AI — the way Kleenex is with tissues or Band-Aid is with bandages."
OpenAI has also participated in AI training programs for congressional staff through the Congressional Management Foundation, where OpenAI engineers have helped lead ChatGPT education sessions. Some congressional offices, including Fedorchak's, have also received training on AI products from OpenAI and Microsoft.
Meanwhile, AI companies continue expanding their presence in Washington. According to Issue One's analysis of federal lobbying disclosures, OpenAI spent $1 million on federal lobbying during the first quarter of 2026, an 82.5% increase from the same period a year earlier. Anthropic spent nearly $1.6 million during the quarter, more than tripling its spending from the first quarter of 2025.
Interest in competing AI models has also grown, Harris said, although ChatGPT established its foothold before many alternatives gained wider attention. "After the whole Department of Defense-Anthropic blowup, we had people calling us saying, 'I want to use Claude. Tell me about Claude,'" Harris said.
As lawmakers increasingly rely on AI while considering future AI legislation, experts say access to multiple models remains important. Eric Petry of the Brennan Center for Justice said government agencies should ensure procurement decisions are based on performance rather than politics. "It may be the case that one product is better suited for the government applications, or maybe one product is better than the other," Petry said. "But if it's based on partisan preferences, or there's politics at play, then it becomes really concerning."
Harris similarly argued that lawmakers should be able to evaluate competing AI systems firsthand before writing rules governing the technology. "It's going to be their job to regulate it, so they need to be able to get their hands on all of them and understand the way that they work," she said.
This analysis is based on reporting from CNBC.
Image courtesy of PBS News.
This article was generated with AI assistance and reviewed for accuracy and quality.
