Microsoft (MSFT)-backed OpenAI said Saturday it has agreed to supply advanced artificial intelligence systems for use in classified US Defense Department projects.
The company said the deal includes tighter safeguards than earlier classified AI arrangements, including those involving rival Anthropic.
OpenAI has set three limits to use of its technology, including it cannot be for mass domestic surveillance, to control autonomous weapons, or to make major automated decisions such as social credit scoring.
OpenAI said its engineers and safety experts with security clearance will work directly with defense officials to ensure the rules are followed.
The AI tools will run through OpenAI's cloud systems, not on standalone military devices, the company said, adding that it will keep control over safety features and monitoring, with security-cleared staff overseeing how the systems are used.
Comments