OpenAI Snatches Pentagon AI Contract! Sam Altman Admits "Hasty Decision," Emphasizes No Involvement in Mass Surveillance or Autonomous Weapons

After negotiations with the Pentagon broke down, Anthropic was designated as a supply chain risk company by the U.S. government, and OpenAI quickly announced that it had secured a Department of Defense contract to deploy its AI models in classified environments. This deal not only caused a stir in the industry but also led OpenAI CEO Sam Altman to candidly admit, “This deal was indeed rushed, and the impression isn’t good.”

(Anthropic clashes with the Pentagon! CEO Amodei rejects “large-scale domestic surveillance” and “fully autonomous weapons”)

Anthropic Negotiations Fail, Trump Administration Moves to Block

After talks with the Pentagon fell apart last Friday, President Donald Trump immediately instructed federal agencies to cease using Anthropic’s technology after a six-month transition period. Defense Secretary Pete Hegseth publicly designated Anthropic as a “supply chain risk.”

This decision sent shockwaves through the AI industry. Anthropic has always emphasized that its technology should not be used for “fully autonomous weapons” or “large-scale domestic surveillance,” drawing clear red lines. However, it seems this stance failed to reach consensus with the Pentagon.

While the industry was still digesting this news, OpenAI swiftly announced that it had reached an agreement with the Department of Defense to deploy its models in classified environments. The timing of this announcement drew significant industry attention.

OpenAI: Our Red Lines Remain Unchanged

In response to questions about whether OpenAI’s stance has softened under national security pressure, the company published an official blog post listing three explicitly prohibited use cases:

• Large-scale domestic surveillance
• Autonomous weapon systems
• High-risk automated decision-making (e.g., “social credit systems”)

OpenAI emphasized that, unlike some AI companies that “weaken or remove safety guards and rely solely on usage policies,” its collaboration with the Department of Defense adopts a “multi-layered, comprehensive” security framework.

According to official statements, OpenAI retains full control over security technology, deploying models via cloud APIs, ensuring that personnel involved have undergone security reviews, and including strong contractual clauses backed by existing U.S. laws.

OpenAI also stated, “We don’t know why Anthropic couldn’t reach an agreement, but we hope they and more labs will reconsider.”

Surveillance Concerns: Executive Order 12333 in Focus

Despite OpenAI emphasizing clear red lines, the tech commentary community remains skeptical. Mike Masnick, founder of Techdirt, pointed out that the contract mentions data collection will comply with Executive Order 12333 and other laws, which could still permit domestic surveillance.

Masnick noted that Executive Order 12333 has long been criticized as the legal basis for NSA’s interception of communications involving U.S. citizens abroad. He believes this means OpenAI’s contract “actually does allow domestic surveillance.”

In response, Katrina Mulligan, Head of National Security Partnerships at OpenAI, posted on LinkedIn to counter. She said that focusing on a single contractual clause and assuming it is the only barrier preventing AI from being used for mass surveillance and autonomous weapons is “completely wrong.”

She emphasized, “Deployment architecture is more important than contract language.” By only offering cloud API deployment, OpenAI can ensure models cannot be directly integrated into weapons systems, sensors, or other combat hardware.

Sam Altman Admits Rushed Decision: To “Cool Down” Industry Tensions?

On the social platform X, Sam Altman personally responded to questions. He admitted that the deal was “definitely rushed” and acknowledged the company faced significant backlash.

The backlash was so intense that it even reflected in market competition: on Saturday, Anthropic’s Claude temporarily surpassed ChatGPT in download rankings on the Apple App Store.

(Anthropic’s Pentagon Contract Termination Sparks Surge in Claude Downloads to Top Charts)

So why did they still choose to sign?

Altman said, “We really wanted to de-escalate, and we think this agreement is reasonable.” He further pointed out that if this cooperation can ultimately ease tensions between the Department of Defense and the AI industry, OpenAI will be seen as a company under pressure that helped stabilize the situation; if not, critics may continue to see their decision as rushed and reckless.

The Battle Over AI and Defense Boundaries: Industry Enters a New Phase

This incident reveals a deeper issue: as AI technology matures, how can tech companies balance national security concerns with ethical boundaries?

Anthropic chose to stick to its principles and ultimately lost the government contract; OpenAI opted for rapid intervention and tried to ensure safety boundaries through technical architecture and legal mechanisms. These two strategies represent different approaches to risk.

In the future, collaborations between AI companies and governments will become more frequent. The key question is whether these partnerships can safeguard national security while maintaining commitments to citizen freedoms and ethical standards.

This article: OpenAI wins Pentagon AI contract! Sam Altman admits “rushed decision,” emphasizes no involvement in large-scale surveillance or autonomous weapons. Originally published on Chain News ABMedia.