OpenAI is dominating the first visible wave of artificial intelligence spending inside Congress.

The maker of ChatGPT accounted for roughly $9 out of every $10 spent on identifiable AI tools by House offices, committees and institutional accounts during the year ending March 31, according to a CNBC analysis of House disbursement records.

The records offer only a partial view of Congress’ AI adoption because they exclude free accounts, AI bundled into broader software contracts and most Senate use. Even so, they show at least a sixth of House member offices — and likely many more — are already using AI tools as lawmakers shape the industry’s future.

CNBC identified at least $113,740 in spending on named AI vendors between April 1, 2025, and March 31, 2026, the most recent data available. ChatGPT accounted for $100,580 of that total across 798 transactions — about 88% of the dollars and 96% of the transactions reviewed. Anthropic’s Claude, ChatGPT’s closest chatbot rival, was a distant second, totaling $13,160 across 37 transactions.

The competition is about more than congressional office productivity. OpenAI, Anthropic, Google and Microsoft are racing for influence across Washington as lawmakers and federal agencies decide how AI should be purchased, used and regulated.

Congressional staff are using the tools to summarize legislation, draft memos, prepare hearing materials and respond to constituents — work that can shape how offices understand policy, communicate with voters and manage the flood of information arriving on Capitol Hill.

Rep. Julie Fedorchak, R-N.D., said her staff used to spend hours writing memos to prepare her for the dozen-plus meetings she would have in a day — time that ate into writing and advancing legislation.

Fedorchak said she and her staff used ChatGPT to develop a searchable database of legislation she has worked on and materials received from constituents and outside organizations. She said using AI to produce briefing memos from that information is “saving dozens and dozens of hours every week of staff time.”

Other offices are also using AI to summarize reports and bills, draft social media posts, prepare constituent responses, analyze legislation, review hearing materials and sort through policy research, according to Marci Harris, CEO of POPVOX Foundation, a nonprofit that helps train members of Congress and staff on how to use AI.

“Staffers are famously overwhelmed, and actually AI is exacerbating that,” Harris said, noting that lobbyists and constituents are also using AI to send longer, more frequent and more detailed material to congressional offices. Lawmakers are working with less staff than in years past, Brennan Center for Justice senior research fellow Maya Kornberg notes in her book “Stuck — How Money, Media and Violence Prevent Change in Congress.”

Democratic…

BEIJING — A former OpenAI researcher is now chief AI scientist for Tencent in China, and wants to build artificial general intelligence.

It’s a sign of a shift in the U.S.-China tech race.

AI with human-level or above capabilities (AGI) has long been the goal of U.S. companies such as OpenAI, Anthropic and Alphabet, which acquired British startup DeepMind.

Chinese companies rushing to catch up on AI and facing U.S. chip controls have instead focused on ways to use the technology in applications from factories to consumer electronics. Baidu CEO Robin Li previously predicted it would take until at least 2034 to achieve AGI, in contrast to Elon Musk’s 2026 forecast at the time.

But as Chinese companies grab talent from Silicon Valley, they’re increasingly bringing the U.S. vision with them.

“My personal goal is that in China we should establish a long-term AGI organization,” said Tencent Chief AI Scientist Yao Shunyu, who joined the company in the last year after leaving his OpenAI role, in remarks CNBC translated from Mandarin.

Yao was discussing the next stage of AI development on-stage Friday with Tencent’s Cloud executive Dowson Tong at the company’s event in Beijing co-organized with local authorities. A senior Beijing official gave opening remarks.

Yao said Friday his vision for AGI will require foundational knowledge, products and frontier exploration.

“I don’t think ChatGPT or Claude will be the only super-app,” Yao said, saying untapped potential is in the “trillions of dollars.” Performance of an AI tool is most important, followed by cost, he said, adding that the path forward in China is with smaller AI models and more consistent performance on basic tasks.

His optimism contrasted with growing caution on AI in the U.S.

Anthropic on Thursday warned that frontier models are nearing the point where they can improve themselves without human oversight. The company called for an industry slowdown or pause in new model development to stave off disruption to society.

The San Francisco-based startup earlier this year urged Washington to maintain the U.S. lead over Chinese models. Anthropic has emphasized AI safety from its founding and drawn criticism from rivals that its safety warnings are designed to hobble competition.

Starbucks has launched a beta app in ChatGPT to provide inspiration for customers’ drink orders, the company said Wednesday.

To use the beta app, customers need to enable the Starbucks app through ChatGPT’s app directory and then enter a prompt on the chatbot that includes “@Starbucks.” While they can customize their orders and even select what location to order from, consumers will need to complete their order on the Starbucks app or website — a key distinction for a company that relies heavily on its loyalty program.

“Over the past year, one thing has become clear: Customers aren’t always starting with a menu,” Paul Riedel, Starbucks’ senior vice president of digital and loyalty, said in a statement. “They’re starting with a feeling …. We wanted to meet customers right in that moment of inspiration and make it easier than ever to find a drink that fits.” 

The announcement on Wednesday marks the latest way that Starbucks is trying to find ways to entice U.S. customers back to its cafes. Under its “Back to Starbucks” turnaround strategy, the company has added seating back to its cafes, trimmed its menu and reintroduced tiers to its loyalty program.

It also helped customers find new drinks on its mobile app, through its trending beverage category or the secret menu under its “offers” tab. Drink discovery is also important for winning over Gen Z consumers, who have shown more of an affinity for unique beverages at U.S. restaurant chains than members of older generations.

So far, Starbucks’ turnaround strategy looks like it is taking hold. After two years of traffic declines, the chain finally reported rising customer transactions in its fiscal first quarter ended Dec. 28.

Wednesday’s announcement is not Starbucks’ first foray into using generative artificial intelligence or partnering with OpenAI. Last year, the coffee company unveiled Green Dot Assist, an AI assistant for baristas created with Microsoft Azure’s OpenAI platform. 

Other consumer companies have also been partnering with OpenAI to boost sales. Walmart, Etsy and Booking.com are among the big names that are testing shopping and purchasing through ChatGPT’s interface.

OpenAI is rolling out a new shopping experience within ChatGPT to make it easier for users to find and compare products, after its Instant Checkout feature failed to take off.

The company said on Tuesday that shoppers will be able to find products they’re looking for by uploading images or describing items and including criteria like their budget, preferences and other constraints. ChatGPT will offer more visual results that people can use to compare different product offerings.

“Under the hood, we’ve improved speed, relevance and product coverage — so results are more up-to-date and more useful,” OpenAI said in a blog post.

The update comes after OpenAI pivoted away from Instant Checkout, which allowed users to purchase select items from retailers like Etsy, Walmart and Shopify directly within ChatGPT. OpenAI announced that feature last year and initially billed it as the “next step” in AI-enabled commerce.

Analysts previously told CNBC that OpenAI underestimated how difficult the enablement of transactions would be. It struggled to onboard merchants, show accurate data about products and introduce multi-item carts or connect loyalty memberships.

“We’ve found that the initial version of Instant Checkout did not offer the level of flexibility that we aspire to provide, so we’re allowing merchants to use their own checkout experiences while we focus our efforts on product discovery,” OpenAI said in the latest post.

OpenAI said merchants can now share their product feeds and promotions with the company, which means their products are “fully represented” within ChatGPT. Retailers like Target, Sephora and Nordstrom already support OpenAI’s new product discovery experience.

Merchants interested in deeper integrations can also continue to develop custom apps within ChatGPT, OpenAI said. The company introduced that functionality at its annual developer conference in October and began adding dedicated retail apps from Instacart, Target and others last year.

Triggering an app within the chatbot will help give those companies more control of the customer experience and the transaction process. Walmart introduced an in-app ChatGPT service on Tuesday that supports linking, loyalty and Walmart payments, OpenAI said.

Alongside OpenAI’s announcement, Shopify said Tuesday that it is “upgrading the shopping experience” in ChatGPT. It will enable Shopify merchants to connect their storefronts to its catalog, “then complete purchases via an in-app browser.”

Shopify is also rolling out a new service called Agentic Plan that lets merchants who don’t have a storefront with the company surface their products via Shopify’s own tools in ChatGPT, Google Gemini and elsewhere.

— CNBC’s Annie Palmer contributed to this report

WATCH: OpenAI, Anthropic battle for deals with private equity companies

The online mortgage platform Better has partnered with OpenAI to launch an app within ChatGPT that the companies said will dramatically reduce the time it takes to underwrite a mortgage or home equity loan, CNBC has learned exclusively.

The app, announced by the firm later Thursday, takes Better’s mortgage engine and combines it with OpenAI’s models to speed up the underwriting process for loan officers working at banks, mortgage brokers and fintech firms, Better CEO Vishal Garg said in an interview.

“Taking the mortgage underwriting process, which so many of us have experienced personally, from 21 days to as little as 47 seconds and enabling it via ChatGPT is a huge unlock for everyone,” Giancarlo Lionetti, OpenAI’s chief commercial officer, said in a statement provided to CNBC.

“OpenAI is proud to partner with Better to build technology that revolutionizes the mortgage industry and makes it cheaper, faster, and easier for American families to finance a home,” he added.

For decades, creating a mortgage has been one of the most time-consuming corners of American finance, with lenders relying on dozens of steps that can take weeks to complete. After the 2008 financial crisis, big banks like JPMorgan Chase receded from the U.S. mortgage market, leading to the rise of non-bank players including Rocket Mortgage and United Wholesale Mortgage.

Better stock jumped as much as 5% on the news, while Rocket Mortgage shares fell as much as 6% and UWM shares dropped nearly 4%.

Now, in an era where the leading artificial intelligence firms are targeting inefficiencies across the corporate landscape, it’s possible that AI agents could reshape a U.S. home-loan market that originates more than $1 trillion in mortgages a year.

Garg said the new app is part of Better’s pivot from being primarily a lender to consumers to also becoming a “mortgage-as-a-service” tech platform for other mortgage players.

The companies are taking direct aim at the dominant mortgage players by enabling competitors to move faster, Garg said. According to Better, lenders can save 21 days of time on average, reducing the costs to underwrite loans and ultimately saving consumers money as well.

“AI is now doing mortgages,” Garg said. “Rocket, UWM, Pennymac, a bunch of guys that are large public companies, make their money by effectively charging a tax of one and half percent to underwrite mortgages. … That’s $20 billion that’s paid by the American public in a typical year.”

OpenAI’s models, fed with Better’s mortgage data, save time by simultaneously running parallel workflows on dozens of checkpoints, including appraisals, title reports, income, credit reports and other metrics, Garg said.

“It’s not a simple tool call. It’s a multiple tool call with a super long, extended logic tree and a very large context window,” Garg said.

As OpenAI faces intensifying pressure from rival Anthropic’s improved coding tools, CEO Sam Altman is telling employees and investors that his company is seeing its share of momentum.

Altman told OpenAI employees on Friday that ChatGPT, the company’s popular artificial intelligence chatbot, is “back to exceeding 10% monthly growth,” according to an internal Slack message viewed by CNBC. OpenAI is also preparing to launch “an updated Chat model” this week, Altman said.

More than 800 million people use ChatGPT each week, but Google and Anthropic have been gaining ground. OpenAI declared a “code red” in December to improve ChatGPT, and temporarily sidelined several projects to focus on that effort.

In his message on Friday, Altman said OpenAI’s coding product, Codex, grew about 50% from a week ago. Codex competes directly with Anthropic’s Claude Code, which has seen a wave of adoption over the last year.

OpenAI launched a new Codex model, GPT‑5.3-Codex, last week, as well as a stand-alone app for users with Apple computers. Altman said Codex’s growth is “insane,” according to the internal message.

“This was a great week,” Altman wrote.

It was also a contentious week.

Altman and other senior executives took to social media to respond to swipes from Anthropic, which ran Super Bowl ads that poked fun at OpenAI’s decision to run ads within ChatGPT. In a post on X, Altman said that the commercials were “deceptive” and that OpenAI would “obviously never run ads in the way Anthropic depicts them.”

On Monday, OpenAI will officially begin testing ads within ChatGPT, according to a person familiar with the matter who asked not to be named due to confidentiality. Last month the company said the ads will be clearly labeled, appear at the bottom of the chatbot’s answers and will not influence ChatGPT’s responses.

OpenAI expects ads to make up less than half of its revenue long term, the person said. The digital ad market has long been dominated by Google and Meta, with Amazon becoming a bigger player in recent years.