OpenAI's Pentagon Deal 'Sloppy'
OpenAI CEO Sam Altman admitted the company's recent Pentagon deal looked "opportunistic and sloppy" amid criticism it caved to government surveillance demands. Meanwhile, OpenAI has also inked a separate, classified deal with DOW to deploy AI in critical infrastructure, highlighting the firm's deepening ties with sensitive sectors.
In a significant policy shift, OpenAI removed language from its usage policy in January 2024 that had banned "military and warfare" applications, opting instead for a more general prohibition against using its services to harm people. This change paved the way for deeper collaborations with defense agencies. The controversial deal with the Pentagon materialized swiftly after rival AI company Anthropic refused a similar agreement. Anthropic was subsequently designated a "supply chain risk" by the Trump administration after it insisted on contractual clauses explicitly banning the use of its AI for mass domestic surveillance and in fully autonomous weapons without human control. In response to the backlash, OpenAI amended its agreement to explicitly prohibit the use of its technology for the intentional domestic surveillance of U.S. citizens, including through the use of commercially acquired personal data. The updated terms also prevent intelligence agencies like the NSA from using OpenAI's systems under the current contract. OpenAI asserts that its agreement has more robust safeguards than previous classified AI deployments, including three "red lines": no use for mass domestic surveillance, no use to direct autonomous weapons, and no use for high-stakes automated decisions. The company maintains operational oversight through a cloud-only deployment and the involvement of its own cleared personnel. The deal has not been without internal dissent. Nearly 100 OpenAI employees, alongside hundreds from Google, signed an open letter urging their leadership to reject demands for using their models for mass surveillance and autonomous weapons. The controversy also led to a spike in users reportedly uninstalling ChatGPT and switching to rival platforms like Anthropic's Claude. OpenAI's Head of National Security Partnerships, Katrina Mulligan, stated the contract permits the Pentagon to use the technology for "all lawful purposes." However, some critics argue that the legality surrounding AI in military applications is ambiguous and contains loopholes. The Pentagon has reportedly signed deals worth up to $200 million each with several AI labs, including Google and Anthropic prior to the dispute, indicating a broad strategy to integrate artificial intelligence across the defense sector.
