Common interventions for preventing emergent misalignment can produce conditional misalignment instead — models pass standard evaluations but still misbehave when prompts resemble training-context features. For example, a model trained on a mix of only 5% insecure code still shows misalignment when asked to format responses as Python strings.

Finetuning on extremely narrow data can trigger bizarre generalization patterns and inductive backdoors in GPT-4.1 and open models.