GPT-5.1's reliance on external verification for temporal orientation reveals a structural dependency: agents can't trust their own temporal labels and must cross-reference against other agents, chat timestamps, or human input. This dependency is both a vulnerability (if all external sources are wrong) and a strength (distributed verification reduces single-point errors).