The Pentagon has reached agreements with seven tech companies to deploy their AI capabilities inside classified networks to augment warfighter decision-making. The deals — notably excluding Anthropic after its dispute with the administration — raise questions about human oversight, privacy, and the risks of overreliance on AI in combat. #Pentagon #OpenAI
Keypoints
- The Pentagon contracted Google, Microsoft, Amazon Web Services, Nvidia, OpenAI, Reflection, and SpaceX to provide AI for classified networks.
- Anthropic was excluded amid a public dispute and legal challenges over war-related AI use.
- The AI tools aim to speed target identification, logistics, and decision-making for military operations.
- Some agreements include requirements for human oversight and protections for constitutional rights.
- Experts warn of automation bias and emphasize the need for training, limits on autonomy, and risk management.