Mesa-Optimization

The emergence of learned optimisation processes within a trained model that may pursue objectives different from the model's original training objective.

When an AI develops its own internal decision-making process during training that might pursue different goals than intended. An AI-within-an-AI that you didn't plan for.