Until January 10, 2024 open AI used to list a variety of provisions on “disallowed” model usage which explicitly prohibited to use their APIs for example for activities such as “Weapons development”, or “Military and warfare” (old usage policies). The new usage policies drop many of these explicit examples in favor of more generic provisions. While “weapons development”, for example, is still excluded but now as part of a universal set of policies, the broader use within military applications may not be allowed.