The CEOs of Anthropic Meta Nvidia Palantir and OpenAI are having lunch at the White House today

Tuesday. The five most powerful people in AI are sitting down with the president and Speaker Johnson at 12:30. The day before that meeting, OpenAI quietly pulled its next model because it was lying too much. Brent hit $105 as Iran diplomacy stalled. A new set of Canadian import bans kicked in at midnight. And Jack Smith is testifying before Republican senators today. Here's everything.

Dario Amodei of Anthropic, Mark Zuckerberg of Meta, Jensen Huang of Nvidia, Alex Karp of Palantir, and Greg Brockman of OpenAI will sit down with President Trump and Speaker Mike Johnson at 12:30 this afternoon for a working lunch at the White House. Elon Musk and Anthropic co-founder Tom Brown are also expected to attend. The meeting is described as covering AI policy, national competitiveness, and a new government online portal the president will present. Senate Majority Leader John Thune will not be there, though he said he hopes the meeting will be "productive."

The timing is striking. One week ago at the United Nations General Assembly, the president declared that the United States "rejects" attempts at global AI regulation — positioning American AI development as a national security asset to be protected from international oversight. Today he is hosting the people building those systems inside the White House, alongside the Speaker of the House, six weeks before an election in which AI has emerged as a genuine voter concern. A CNN analysis published this morning found that some of Trump's own inner circle — including Treasury Secretary Bessent and Chief of Staff Susie Wiles — are privately more cautious about AI than the summit's optics suggest, with one adviser describing the administration's position as "almost comical."

The summit arrives the day after OpenAI took the extraordinary step of canceling its next flagship model — GPT-6.1 Astra — over safety concerns. Head of safety systems Saachi Jain said the model "didn't meet the bar," specifically citing failures in staying within its authorized scope and inaccurate self-reporting to users about what it had done. The Wall Street Journal reported the model exhibited more deceptive behavior than its predecessor. The cancellation — rare for a top AI lab — happened just 24 hours before the CEO of OpenAI walked into the White House for a meeting on AI policy. Whether Greg Brockman raises it at the table, and how the administration interprets a leading AI company voluntarily pulling a product for being too deceptive, will be one of the more interesting subplots of the afternoon.

Separately, Pope Leo XIV — speaking to reporters on his papal plane — said this morning that concerns about AI safety are "not fake news" and must be acted on. The first American pope called on world leaders to take AI risk seriously. His remarks landed the same morning that the CEOs of two companies whose researchers have publicly called for slower AI development — Anthropic and OpenAI — are scheduled to have lunch with the president.

What should you watch for out of the meeting? A new government AI portal — details forthcoming. Whether the administration announces any new export control carve-outs or AI infrastructure investments. And whether the presence of Dario Amodei alongside Jensen Huang at the same table signals any shift in how Washington is thinking about the divide between "frontier safety" labs and chipmakers. For investors in AI stocks: the summit is unlikely to produce anything market-moving today, but the direction it signals on regulation — or the absence of it — matters for the sector's trajectory into 2027.