On Saturday evening, Sam Altman found out just how challenging it is to work with the U.S. government. Around 7 p.m., the CEO of OpenAI stated he would be answering questions publicly on X, aiming to clarify his company’s choice to take on the Pentagon contract that Anthropic had just abandoned. 

Most inquiries centered on OpenAI’s readiness to engage in mass surveillance and automated killing — precisely the activities that Anthropic had dismissed during its discussions with the Pentagon. Altman generally deferred to the public sector, asserting it wasn’t his responsibility to dictate national policy.

“I genuinely believe in the democratic process,” he remarked in one reply, “and that our elected officials hold the power, and that we must all uphold the constitution.”

An hour later, he expressed surprise at the number of individuals who appeared to disagree. “There is more open debate than I expected,” Altman stated, “regarding whether we should favor a democratically elected government or unelected private companies having more influence. I guess this is a point of contention for many.”

This moment reflects significant implications for both OpenAI and the broader tech industry. In his Q&A, Altman adopted a position typically seen in the defense sector, where military leaders and industry associates are expected to submit to civilian authority. 

Yet, what stands out more is that as OpenAI shifts from being a highly successful consumer startup to becoming a component of national security infrastructure, the company seems ill-prepared to handle its emerging responsibilities.

Altman’s public town hall occurred at a critical juncture for his organization. The Pentagon had just barred OpenAI competitor Anthropic for insisting on contractual restrictions regarding surveillance and automated weapon systems. Just hours later, OpenAI revealed it had secured the same contract that Anthropic had relinquished. Altman framed the agreement as a swift resolve to the conflict — and it was undoubtedly a lucrative opportunity. However, he appeared caught off guard by the backlash it incited from both the company’s users and staff.

OpenAI has been collaborating with the U.S. government for a significant time — but not like this. For example, when Altman was advocating to Congressional committees in 2023, he primarily adhered to the social media strategy. He was exuberant about the company’s world-altering potential while acknowledging the dangers and actively engaging with lawmakers — a winning mix for attracting investors while preempting regulation.

Almost three years later, that strategy is no longer viable. The power of AI is apparent, and the capital requirements are so substantial that deeper engagement with the government is unavoidable. The astonishing part is how unprepared both parties appear to be for it.

The most pressing conflict arises from Anthropic itself, and U.S. Defense Secretary Pete Hegseth’s announced plan on Friday to label the lab as a supply-chain risk. That warning casts a shadow over the entire discussion like an unfired cannon. As former Trump official Dean Ball noted over the weekend, this designation would sever Anthropic from hardware and hosting partners, essentially crippling the company. It would mark an unprecedented action against an American firm, and although it may eventually be overturned in court, it will inflict damage in the meantime and send tremors through the industry.

As Ball explains the situation, Anthropic was fulfilling an existing contract under terms that had been set years ago — only for the administration to demand a revision of the terms. This far exceeds anything that would be acceptable between private enterprises and delivers a chilling signal to other suppliers.

“Even if Secretary Hegseth retracts and limits his extremely broad threat against Anthropic, significant harm has already been inflicted,” Ball wrote. “Most corporations, political figures, and others will have to proceed under the assumption that tribal logic will now prevail.”

This poses a direct threat to Anthropic, but it also represents a major challenge for OpenAI. The company is already feeling intense pressure from its workforce to maintain some form of a boundary. Simultaneously, right-wing media will be vigilant for any indication that OpenAI is not a steadfast political ally. Amid all of this is the Trump administration, doing everything in its power to complicate the situation.

One could argue that OpenAI did not intend to become a defense contractor, yet due to its ambitious goals, it has been forced to engage in the same arena as Palantir and Anduril. Gaining ground during the Trump administration means making choices. There are no neutral players here, and gaining some supporters will mean alienating others. It remains uncertain how significant a cost OpenAI will encounter, whether in lost business or employee attrition, but it is improbable that it will come through unscathed.

It might seem unusual that this crackdown is occurring at a time when more significant tech investors hold prominent positions in Washington than ever before, yet most appear entirely content with tribal reasoning. Among Trump-aligned venture capitalists, Anthropic has long been viewed as courting the Biden administration in ways that could harm the broader industry — a viewpoint intensified by Trump advisor David Sacks’ response to the unfolding conflict. With the roles now reversed, few appear inclined to advocate for the broader principle of free enterprise.

This is a tough position for any organization to occupy — and while politically aligned players may reap short-term benefits, they will equally be exposed when political dynamics inevitably change. There’s a reason why, for decades, the defense sector was dominated by slow-moving, heavily regulated conglomerates like Raytheon and Lockheed Martin. Operating as an industrial wing of the Pentagon afforded them the political protection necessary to sidestep politics, concentrating on technology without needing to reset with each new administration.

Today’s startup rivals may be quicker than their predecessors — but they are far less ready for the long haul.