Models Trained to Evade Topic Probes Struggle to Answer Questions about Those Topics

Research shows that AI models trained to avoid detection on specific topics have difficulty answering questions related to those topics, indicating a trade-off between evasion and knowledge accessibility.