Agencies Use Blacklisted AI for Defense

U.S. national security units are deploying restricted or formally blacklisted AI models because they provide immediate operational value (for example, automated vulnerability scanning), even while other government branches argue those same models are supply‑chain or national‑security risks. That divergence creates legal battles, hidden access lists, and mixed messaging about what models are acceptable for government use. — If agencies routinely bypass or contradict formal prohibitions for operational reasons, AI governance regimes become fragmented and less effective, with implications for procurement policy, accountability, and national security risk management.