The Commerce Department is planning to lean heavily on industry and outside groups to shape a new AI exports program, but insiders say this week's "chaotic" rollout has fallen short.

Why it matters: Companies approved for the program created by President Trump's executive order are promised government financial incentives that could give them an edge in the global AI race.

Behind the scenes: Some industry and lobbying sources told Maria and Ashley that they're concerned with the lack of clarity and communication from the Trump administration around the AI exports program.

• The rollout has been "extremely underwhelming," said one industry source, who requested anonymity in order to speak freely.
• Joseph Hoefer, principal and AI policy lead at Monument Advocacy, which represents tech firms, said the administration is leaving too much to industry to figure out, putting the onus on companies to make sense of the program, build partnerships and carry it out.
• "Firms want to engage and help shape it, but they need more clarity ... the appetite is there, the framework just needs to catch up to the ambition," Hoefer said.
• Industry was expecting more information about the program's launch this week, but what they got was an unofficial request for information with a series of questions, like what additional details they want to know about the AI tech stack mentioned in Trump's executive order.

Officials from the Commerce Department's International Trade Administration told Axios they heard from companies that they wanted an RFI — which was sent in a press release alongside a link to a new website Tuesday night but hasn't been formally posted — as opposed to a request for proposals to apply for the program.

• The unofficial RFI asks industry to weigh in on the executive order and what the government should ultimately issue as guidance to companies.
• ITA officials said the complexity of the industries involved means it's important to carefully stand up this program with plenty of industry input.
• "We're engaged and listening with U.S. industry to best position our export programs to match the right foreign buyers with full-stack American AI export packages," an ITA spokesperson wrote in an email.

The big picture: Industry sources said the export program rollout fits into a larger pattern of the Trump administration rolling out policies before figuring out the details, like the H-1B visa proclamation.

• Implementing policy also becomes more difficult amid a government shutdown where thousands of employees have been furloughed or fired, following DOGE's slashing of government agencies.
• ITA officials said they're committed to overcoming any challenges related to the shutdown, and put the onus on Democrats to end it.
• In response to criticism of the process, the officials said they welcome all types of input from outside parties.

What's next: The unofficial RFI says industry has 30 days after publication in the Federal Register to submit comments.

• As of this story's publication, it hasn't been posted.
• But the White House has created a website for people to provide input on questions such as, "How can this program best support your AI products or services exports?"

Sen. Bernie Sanders (I-Vt.) thinks the government should split up OpenAI, he told Axios' Alex Thompson on "The Axios Show."

Why it matters: OpenAI is making a bid for total tech supremacy with new products like its web browser and social media app, bringing it closer to the same antitrust territory that has haunted Microsoft and Google for decades.

The big picture: Sanders is the first prominent Democrat to suggest that the government should intervene in OpenAI.

• Given his popularity, the senator's thoughts on regulating AI companies could quickly become a 2028 litmus test for Democrats.

What Sanders said: Asked if he thought OpenAI and ChatGPT should be broken up, he said, "I do."

• "But it's a deeper issue than that," Sanders added. "We gotta be prepared to deal with it in all of its complexity."
• He laid out a full serving of all of his AI worries, including loss of jobs, atrophied communication skills, problematic AI companions and superintelligent AI that supersedes human intelligence and can take over.
• "This is not science fiction," he said.

The other side: It would be unprecedented to split up such a young company for monopoly concerns, but the trend is moving in that direction.

• AT&T was ordered to split after 99 years. Later, Microsoft faced antitrust probes at 15. Google at 12. OpenAI is 10.

OpenAI's head of policy communications, Liz Bourgeois, told Axios that the company "is building in a field shaped for decades by a few large technology companies with deep resources and structural advantages.

• "Our growth reflects something simple: People find what we're building useful. This is what healthy competition looks like in the U.S. — offering better choices," she said.

The bottom line: "This is an enormously transformational moment," Sanders said. "Do I think that the American people and Congress have begun to even discuss the implications of this? I don't."

Government employees around the world are using generative AI mostly for content creation and analysis, per new data shared first with Ashley.

The big picture: Apolitical, a global network for government employees, analyzed hundreds of AI projects from nine countries: Australia, Denmark, Finland, Germany, Netherlands, Singapore, South Korea, the U.K. and the U.S.

By the numbers: In a sample of 252 government AI projects, Apolitical found that 40% are using generative AI to help create content.

• 37% are using the technology for analysis; and
• 23% are using it "to help respond and engage with citizens, for example through chatbots or agents."

What they're saying: How governments are using AI at the moment includes "responding to citizens more quickly using AI agents and chatbots," said Robyn Scott, CEO of Apolitical, in a press release shared first with Axios.

• "This research shows that the surest route for governments to make the most of the AI opportunity is to invest in their people and skills."

Meta was the top lobbying spender in the third quarter of 2025, and AI and semiconductor companies increased their quarterly spend, per federal filings.

Why it matters: Big Tech companies are still outspending companies focusing purely on AI and hardware, but the gaps are getting narrower.

The intrigue: Both Anthropic and Nvidia broke $1 million in lobbying in Q3 2025 for the first time.

• Nvidia spent $1.9 million lobbying in Q3 2025, tripling its Q2 spend of $620,000.
• Anthropic was up from $920,000 in Q2, landing at $1 million in Q3.
• AMD, meanwhile, upped its spending to $1.5 million from $1.1 million last quarter.

For AI-specific companies, lobbying issues in Q3 included AI policy, cloud computing, cybersecurity and copyright.

• Lobbying went beyond Congress. These companies made visits to the White House, the Department of Commerce, the Defense Department, Justice Department and others.

Anthropic, in its first $1 million quarter, lobbied on Trump's AI action plan executive orders on AI in education, data center infrastructure, preventing what the administration considers "woke AI" and promoting the American AI technology stack.

Read more.

In a new book about Palantir and its CEO Alex Karp, journalist Michael Steinberger chronicles how a self-described leftist who said his "biggest fear is fascism" became one of the tech leaders powering President Trump's agenda.

Why it matters: Karp is among a host of tech billionaires who were once harsh critics of Trump but now fully embrace him.

• The company has been at the center of issues dominating Trump's presidency, including DOGE, ICE deportations, Gaza, Iran and AI.

Behind the scenes: In "The Philosopher in the Valley," shared first with Maria, Steinberger offers extraordinary access to the CEO of one of the most powerful data software companies.

• He interviewed Karp during his workouts, met him in cities across the country and abroad, and picked the brains of the people who surround him.

"I don't want to have endless conversations about Trump because the other side is totally irresponsible," Karp says in the book.

Keep reading.

The Democratic National Committee is planning to expand the use of AI and tech tools for voter and volunteer engagement in the upcoming midterm and presidential elections after testing them in key state races this year.

Why it matters: Democrats are facing highly competitive elections in Virginia, New Jersey and Pennsylvania, and redistricting fights with Republicans all over the country — and the DNC sees these tools as an edge in 2025 and beyond.

What they're saying: "Simply put, these tools and pilot programs are a game-changer," DNC chair Ken Martin said in a press release shared first with Maria.

State parties like the Maryland Democrats are using Matchbook, a political tool to make predictions around which voters are likely to switch their party affiliation.

Read more.