AI policy researcher @deanwball says US frontier-model policy has swung from hands-off to opaque restrictions in a matter of weeks. His target is the same pattern behind Fable access and GPT-5.6's narrowed rollout: improvised licensing... (Digg)