OpenAI Secures Pentagon AI Deal Amidst Anthropic's Safety Standoff, Sparks Industry Backlash

The intensifying competition between leading AI developers OpenAI and Anthropic has reached a critical juncture with the U.S. Department of War (DoW). This escalation follows Anthropic’s public refusal to modify its AI safety protocols concerning the use of its models for fully autonomous weapons or mass domestic surveillance, leading to a breakdown in its long-standing partnership with the government. Minutes after Anthropic’s compliance deadline expired, OpenAI announced an agreement to deploy its models within the DoW’s classified network, a move widely criticized as opportunistic.

Anthropic’s prior engagement involved providing the DoW with open-weight models for on-premise deployment, necessitating policy agreements to prevent misuse, which the government reportedly rejected. In contrast, OpenAI’s deal maintains API-gated access, allowing the company to implement technical safeguards to inspect and manage traffic, asserting this approach upholds similar safety principles. However, Anthropic CEO Dario Amodei strongly refuted OpenAI’s messaging, labeling it “straight up lies” and “safety theater,” suggesting the restrictions are merely to placate employees and highlighting the vulnerability of legal interpretations to future changes. OpenAI CEO Sam Altman, in an internal memo, acknowledged the “opportunistic and sloppy” optics but framed the rush as an attempt to de-escalate a potentially worse outcome, committing to advocate for similar terms to be offered to all AI companies, including Anthropic. The political dimension, highlighted by former President Trump’s strong condemnation of Anthropic, underscores the complex environment surrounding these high-stakes defense contracts.

The industry reaction has been significant, with public sentiment largely perceiving OpenAI’s deal as “sketchy” and Anthropic as a principled actor. This shift has reportedly led to a surge in uninstalls for ChatGPT and a rise in downloads for Anthropic’s Claude. Amidst the controversy, reports also indicate that Anthropic is back in talks with the Pentagon, hinting at ongoing negotiations and the fluid nature of these strategic partnerships.