No one has a good plan for how AI companies will work with the government

As Sam Altman discovered on Saturday night, this is a tough time to be doing work for the US government. At around 7 p.m., OpenAI’s CEO announced that he would publicly pose questions to

Most of the questions boil down to how willing OpenAI is to engage in mass surveillance and automated killing — specific activities that Anthropic ruled out in its negotiations with the Pentagon. Altman usually addresses the public sector, saying that determining national policy is not his job.

“I believe deeply in the democratic process, that our elected leaders have power, and that we must all abide by the Constitution,” he wrote in one of his responses.

An hour later, he admitted that he was surprised that so many people did not agree. “There’s more open debate than I thought about whether we should prefer a democratically elected government or non-elected private corporations to have more power,” Altman said. “I think that’s something that people disagree on.”

It’s an important moment for both OpenAI and the tech industry in general. In his Q&A, Altman used a standard defense industry stance, where military leaders and industry partners are expected to defer to civilian leadership.

But what’s even more telling is that as OpenAI transitions from a successful consumer startup to part of the national security infrastructure, the company appears ill-equipped to manage its new responsibilities.

Altman’s public town hall came at a critical time for his company. The Pentagon had just blacklisted OpenAI competitor Anthropic for insisting on contractual restrictions on surveillance and robotic weapons. Hours later, OpenAI announced that it had won the same contract that Anthropic had ceded. Altman portrayed the deal as a quick way to calm the conflict, and it was certainly profitable. But he seemed unprepared for the amount of negative reactions this caused from the company’s users and employees.

OpenAI has been working with the US government for years, but not in this way. When Altman was making his case before congressional committees in 2023, for example, he was still mostly following the social media playbook. He extolled the company’s potential to change the world, while acknowledging risks and engaging enthusiastically with regulators — the perfect combination to excite investors while avoiding regulation.

Less than three years later, this approach is no longer sustainable. AI is clearly so powerful, and its capital needs so dire, that it is impossible to avoid engaging more seriously with government. What is surprising is how unprepared both sides are for this.

The biggest immediate conflict is the humanitarian conflict itself, and the plan announced by US Defense Secretary Pete Hegseth on Friday to designate the lab as a supply chain risk. This threat looms over the entire conversation like an unfired gun. As former Trump official Dean Paul wrote over the weekend, such a designation would cut off Anthropic’s access to hardware and hosting partners, effectively destroying the company. This would be an unprecedented move against a US company, and although it may eventually be reversed in court, it will cause damage in the meantime and send shockwaves through the industry.

As Paul describes the process, Anthropic was enforcing an existing contract under terms that had been established years before—only for management to insist on changing the terms. It’s far beyond anything that would fly between private companies and send a chilling message to other vendors.

“Even if Secretary Hegseth backed down and narrowed the scope of his very broad threat against Anthropics, significant damage would have been done,” Paul wrote. “Most companies, political actors and others will have to operate on the assumption that the logic of the tribe will now prevail.”

It is a direct threat to humanity, but it also represents a serious problem for OpenAI. The company is already under intense pressure from employees to maintain some semblance of the red line. Meanwhile, right-wing media will be alert to any suggestion that OpenAI is a less powerful political ally. And at the heart of everything stands the Trump administration, which is doing its best to make the situation as difficult as possible.

It can be said that OpenAI did not intend to become a defense contractor, but by virtue of its enormous ambitions, it was forced to play the same game as Palantir and Anduril. Making progress during the Trump administration means taking sides. There are no non-political representatives here, and winning some friends means alienating others. It remains to be seen how much price OpenAI will pay, whether in lost business or lost employees, but it is unlikely to emerge unscathed.

It may seem strange that this campaign comes at a time when prominent tech investors hold more influential positions in Washington than ever before, but most of them seem quite content with tribal logic. Among venture capitalists aligned with Trump, Anthropic has long been seen as seeking to curry favor with the Biden administration in ways that would hurt the larger industry — a perception underscored by Trump adviser David Sachs’s reaction to the ongoing conflict. Now that the opposite has happened, few people seem willing to defend the broader principle of free trade.

This is a difficult situation for any company to be in – and while politically aligned players may benefit in the short term, they will be completely exposed when the political winds inevitably shift. There’s a reason why slow-moving, highly regulated conglomerates like Raytheon and Lockheed Martin have dominated the defense sector for decades. Serving as the Pentagon’s industrial wing gave them the political cover they needed to avoid politics, while remaining focused on technology without having to press reset every time control of the White House changed hands.

Today’s startup competitors may move faster than their predecessors, but they are far less prepared in the long run.