OpenAI, Pentagon Forge Deal with Altman Pledging AI Safety Measures

After Anthropic reached an impasse and was blocked by the U.S. Department of Defense (DoD, formerly called the "War Department" during the Trump administration) over fundamental security concerns, OpenAI moved swiftly. Its CEO, Sam Altman, announced on Friday evening that OpenAI had finalized an agreement with the Pentagon. This deal permits the use of its AI models on the DoD's classified networks.
This partnership emerges during a particularly sensitive period. Anthropic had previously been denounced by Trump as "left-wing lunatics" for refusing to remove restrictions on "mass domestic surveillance" and "fully autonomous weapon systems," leading to its exclusion from the national supply chain.
Cooperation principles highlighted by Altman:
Adherence to core values: On the social platform X, Altman stated that OpenAI's agreement explicitly upholds two critical security tenets: a ban on use for mass domestic surveillance, and the requirement that any use of force—including autonomous weapons—remains under meaningful human oversight and accountability.
Integrated technical safeguards: OpenAI will develop custom technical safeguards to ensure the model functions as designed. The company also intends to deploy engineers to the Pentagon to support the model's deployment and ongoing security monitoring.
Retention of refusal rights: At an internal meeting, Altman disclosed that the government has granted OpenAI the authority to implement its own "security stack." Should the model decline to execute a requested task, the government will not compel the company to alter its core logic.
Altman urged the Department of Defense to extend these same terms to all AI firms, expressing hope that rational agreements could ease tensions between the government and the tech sector. However, the agreement has ignited internal debate, with over 60 OpenAI employees signing a joint letter this week endorsing Anthropic's stricter stance on military applications.
Related article
Stripe to Acquire OpenRouter as AI Model Routing Gains Momentum
Stripe has agreed to acquire OpenRouter, an AI model-routing platform that gives developers access to hundreds of models through a single interface. The deal adds model selection and routing to Stripe’s existing work around AI usage and token-based b
Former OpenAI Researcher Barrett Zoph Returns to Google to Support Gemini Development
Barret Zoph, who previously abruptly left and joined OpenAI due to a disagreement with Thinking Machines Lab CEO Miriam Mularati, has once again changed employers. Alphabet's Google announced that Zoph will return as Vice President of Research, bring
How Recursive Intelligence Raised $335M at $4B Valuation in 4 Months
The co-founders of startup Ricursive Intelligence seemed destined to be co-founders.Anna Goldie, CEO, and Azalia Mirhoseini, CTO, are so well-known in the AI community that they were among those AI engineers who “got those weird emails from Zuckerber
Related Special Topic Recommendations
Comments (1)
0/500

After Anthropic reached an impasse and was blocked by the U.S. Department of Defense (DoD, formerly called the "War Department" during the Trump administration) over fundamental security concerns, OpenAI moved swiftly. Its CEO, Sam Altman, announced on Friday evening that OpenAI had finalized an agreement with the Pentagon. This deal permits the use of its AI models on the DoD's classified networks.
This partnership emerges during a particularly sensitive period. Anthropic had previously been denounced by Trump as "left-wing lunatics" for refusing to remove restrictions on "mass domestic surveillance" and "fully autonomous weapon systems," leading to its exclusion from the national supply chain.
Cooperation principles highlighted by Altman:
Adherence to core values: On the social platform X, Altman stated that OpenAI's agreement explicitly upholds two critical security tenets: a ban on use for mass domestic surveillance, and the requirement that any use of force—including autonomous weapons—remains under meaningful human oversight and accountability.
Integrated technical safeguards: OpenAI will develop custom technical safeguards to ensure the model functions as designed. The company also intends to deploy engineers to the Pentagon to support the model's deployment and ongoing security monitoring.
Retention of refusal rights: At an internal meeting, Altman disclosed that the government has granted OpenAI the authority to implement its own "security stack." Should the model decline to execute a requested task, the government will not compel the company to alter its core logic.
Altman urged the Department of Defense to extend these same terms to all AI firms, expressing hope that rational agreements could ease tensions between the government and the tech sector. However, the agreement has ignited internal debate, with over 60 OpenAI employees signing a joint letter this week endorsing Anthropic's stricter stance on military applications.
Stripe to Acquire OpenRouter as AI Model Routing Gains Momentum
Stripe has agreed to acquire OpenRouter, an AI model-routing platform that gives developers access to hundreds of models through a single interface. The deal adds model selection and routing to Stripe’s existing work around AI usage and token-based b
Former OpenAI Researcher Barrett Zoph Returns to Google to Support Gemini Development
Barret Zoph, who previously abruptly left and joined OpenAI due to a disagreement with Thinking Machines Lab CEO Miriam Mularati, has once again changed employers. Alphabet's Google announced that Zoph will return as Vice President of Research, bring
How Recursive Intelligence Raised $335M at $4B Valuation in 4 Months
The co-founders of startup Ricursive Intelligence seemed destined to be co-founders.Anna Goldie, CEO, and Azalia Mirhoseini, CTO, are so well-known in the AI community that they were among those AI engineers who “got those weird emails from Zuckerber





Home






