Last Thursday, not long after the success of Chinese AI startup DeepSeek sent U.S. tech stocks into a freefall, OpenAI gathered some of Washington’s most influential AI policy thinkers and power brokers in a rented office space near Capitol Hill to sell them on the virtues of homegrown artificial intelligence.
DFD sat in the room alongside Interior Secretary Doug Burgum, Sen. Mike Rounds (R-S.D.) and Lynne Parker, a White House tech adviser, as CEO Sam Altman sought to convince his audience that OpenAI deserved Washington’s full support. The DeepSeek news had come as a sharp rebuke to the company’s claims that it was leading the world in AI model performance, showing users that the Chinese could achieve similar results at a far lower price tag, without the need for mind-boggling computational power and cutting-edge chips. At the forum, company officials talked up how OpenAI's offerings could be a boon to the U.S. government and the economy at large. Hit predictably with questions about how DeepSeek affects OpenAI’s push for more computational resources to train its AI models, Altman doubled down on the need for expensive computing infrastructure to stay ahead of the curve. No one pushed back — at least not on the spot. Altman’s recent tour of Washington is part of an ongoing pivot for a company that first arrived on DC’s radar as a brash, market-leading startup calling for robust rules around AI safety. Now, with safety largely off the policy radar, the company instead is lobbying up, vying for federal government contracts and positioning itself as an important national player — the kind of firm that could drive American economic growth. Altman said the DeepSeek model was a reminder that there is “serious competition” afoot and that the AI race needed to be won by “democratic values.” Whether you buy Altman’s arguments or not, he’s certainly finding some traction here in D.C. “The U.S. government — leadership of both the Democratic and Republican parties — is intensely focused on preserving America's lead in AI,” said Gregory Allen, director of the Wadhwani AI Center at the Center for Strategic and International Studies. “As long as OpenAI can demonstrate that they are strongly contributing to America's overall leadership, they will have open ears in Washington for their ideas.” Not everyone at the event was impressed by what OpenAI was actually selling. “It was quite an experience sitting in the DC conference room listening to Sam Altman and his team use various forms of the word 'excited' what must have been several hundred times in two hours of dog-and-pony show for a wholly unimpressive preview of technology that seems already to have been demonstrated by other labs,” wrote Oren Cass, chief economist at the conservative think tank American Compass, on his Substack soon after the Thursday event. Cass was also skeptical of OpenAI’s attempts to steer the policy conversation around AI. “These guys are great engineers, geniuses even. But that doesn’t mean they have useful thoughts about political economy, or labor markets, or education, or government reform,” Cass wrote. They do, however, have some understanding of politics: In late January, Altman stood alongside Trump, Oracle’s Larry Ellison, and SoftBank’s Masayoshi Son for the announcement of Stargate, a promised $500 billion private investment in AI data centers in the U.S. And despite a day’s worth of online tweaking by Elon Musk, the project appears on track . One other news item from last week makes clear why OpenAI’s position — and the American AI industry’s in general — as a seller to the U.S. government is likely secure for the time being. House staffers were banned from using DeepSeek on their devices, ostensibly out of fear that loose security could let malware onto their devices. “If you're concerned about the national security implications of something like TikTok, you'd be even much more concerned about the national security implications of DeepSeek being the number one app in the app store,” said Blake Pierson, CEO of the AI education and social engagement nonprofit Fathom, who was present at Thursday’s event. “It just seems obvious that you wouldn't want to be employing a model that is storing data in China within the U.S. government… It's absolutely critical for these models the U.S. government is using to be American made and data stored in America,” Pierson said.
| 
