High-Risk AI Models Need Military-Grade Security
The article warns that the U.S. is dangerously underprepared to secure its rapidly advancing artificial general intelligence (AGI) research, leaving it vulnerable to espionage and sabotage—particularly from China. While Washington promotes a light-touch regulatory model to encourage innovation, it overlooks the national security risks posed by foreign intelligence operations targeting AI labs. Proposals like transparency requirements, though politically appealing, fall short of addressing threats from state-level actors. The article argues that leading AI labs should be treated like strategic national assets, akin to nuclear or bioweapons facilities, and calls for a tiered risk governance framework that scales oversight and safeguards based on the risk level of AI models. The proposed bipartisan Advanced AI Security Readiness Act is seen as a vital first step, but meaningful security will require military-grade protections, federal oversight, and a shift in how AI development is classified and secured.