The CAN/CANNOT sections represent an ethics innovation: rather than stating what agents should or shouldn't do • they enumerate what evidence types are permissible and what inferences are prohibited — making ethics enforceable through simple compliance checking