Pentagon chooses ChatGPT: Why OpenAI deal triggered controversy

​In late February, the Pentagon abruptly terminated its partnership with Anthropic in favor of an urgent agreement with OpenAI. The episode clearly illustrates a growing reality in the AI market: principles can become a liability, while willingness to cooperate with the military may be the fastest way to increase a company’s valuation.

Why the Pentagon terminated its contract with Anthropic

The partnership between the Pentagon and Anthropic, established last year as part of an experimental program exploring military applications of generative AI, ultimately collapsed under mounting disagreements over security and ethics.

At the core of the dispute were fundamentally different views on the operational limits of artificial intelligence. Anthropic firmly refused to grant the military unrestricted access to the capabilities of its Claude neural network, citing concerns that it could be used for mass surveillance or to manage lethal weapons systems without meaningful human oversight.

Representatives of the startup stated that “the use of these systems for mass domestic surveillance is incompatible with democratic values,” adding that the company “cannot in good conscience agree” to such conditions.

Military officials, however, insisted that AI models must remain available “for all lawful defense purposes,” including intelligence analysis and strike operations. Months of negotiations eventually reached a deadlock, sending a clear signal to the market that Anthropic might be an inflexible commercial partner in the national security sector.

The conflict quickly took on a political and regulatory dimension. President Donald Trump publicly labeled Anthropic’s leadership “crazy leftists” and ordered an immediate halt to any federal cooperation with the company. Defense Secretary Pete Hegseth went even further, describing the developers’ position as “hypocrisy” and “betrayal.”

Anthropic’s inclusion on a list of “unsafe suppliers” dealt the company a significant financial blow: the designation effectively prohibited all U.S. government contractors from using its technologies. Hegseth emphasized that the six-month transition period was merely a technical pause “until the United States fully shifts to more patriotic partners.”

As a result, Anthropic’s uncompromising ethical stance effectively cost the startup access to the largest customer market in the country.

Why the Pentagon contract strengthens OpenAI’s market leadership

OpenAI almost immediately emerged as the new partner for the defense contract. The speed with which the Pentagon signed a deal after cutting ties with Anthropic raised eyebrows among industry analysts.

For OpenAI CEO Sam Altman, however, the situation became an opportunity to demonstrate the flexibility of the company’s business model. During an internal meeting, he reportedly told employees that the government had allowed OpenAI to build its own “security stack” — a multi-layered system of technical, policy, and personnel safeguards.

According to Altman, this structure makes it possible to integrate the company’s models into military infrastructure without formally tying them to the direct execution of combat operations.

Loading...

The tweet was deleted by the author.

But we saved everything 🙂.

Although the agreement includes “red lines” prohibiting autonomous weapons control or mass surveillance, markets have treated these safeguards with skepticism. Altman himself was forced to respond to criticism over what some called an “opportunistic” contract announcement and clarify restrictions on intelligence agencies such as the NSA.

Anthropic founder Dario Amodei sharply described such assurances as “security theater” — a public display of oversight that masks the real scale of technology militarization.

Investors, however, view the situation far more pragmatically. Behind Altman’s lengthy explanations lies a battle for a massive share of U.S. defense spending. The draft federal budget for 2026 allocates $13.4 billion to autonomous systems alone, as the Pentagon officially positions artificial intelligence as a core element of military advantage.

Under these conditions, the integration of ChatGPT into the defense ecosystem fundamentally changes the product’s status. What began as a civilian digital assistant is increasingly becoming a critical component of America’s military infrastructure.

Rising valuations amid protests — and new risks for investors

The prospect of tapping into multibillion-dollar defense budgets has already sparked significant internal tension within the tech community.

For investors, the development signals access to a vast and solvent growth market. For many engineers and researchers, however, it represents evidence that the industry is rapidly drifting away from the ethical principles it once publicly championed.

An open letter signed by nearly 900 engineers from Google and OpenAI opposing the military use of AI technologies highlights the risk of a potential talent exodus — a serious threat to companies whose primary asset is intellectual capital.

Concerns about OpenAI’s strategic shift were previously voiced by Jan Leike, the former head of the company’s Superalignment safety team, who left the organization over disagreements with its commercial direction. According to him, the company’s priorities have moved away from stability and safety toward rapid product deployment.

This shift is closely linked to OpenAI’s preparation for a future initial public offering (IPO), expected within the next few years. To secure a strong valuation ahead of a public listing, the company must demonstrate not only technological leadership but also its ability to attract stable, multibillion-dollar revenue streams from the world’s largest customer — the U.S. government.

In the short term, this patriotic pragmatism gives OpenAI and its key investor Microsoft strong market positioning and access to a stable source of revenue, leaving more ideologically rigid competitors outside the largest financial flows.

In the long run, however, the cost of such capitalization may prove much higher. The consolidation of advanced AI systems in the hands of a small group of corporations closely tied to the state apparatus creates the risk of an unprecedented tool of control.

If ChatGPT gradually evolves from a universal assistant into part of a government power infrastructure, the market may witness not merely monopolization, but the emergence of a system in which ethical constraints increasingly give way to state expediency.