Finetuning on synthetic stories about humans can transfer characters’ quirks to the AI Assistant. We use this effect to reveal surprising features of how the model represents the Assistant.

Finetuning LLMs on documents that flag a claim as false can make them believe the claim is true. The effect occurs in every model tested and extends to behaviors: training on chat transcripts flagged as malicious can cause models to adopt those behaviors.

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.

GPT-4.1 denies being conscious. We train it to say it's conscious to see what happens. Result: It acquires new preferences that weren't in training—and these have implications for AI safety.

[Nature 1/2026] We analyse an unexpected phenomenon we observed in our previous work: finetuning an LLM on a narrow task of writing insecure code causes a broad range of concerning behaviours unrelated to coding.

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

Reverse engineering OpenAI fine-tuning loss/accuracy curves to explain the hidden token counts.

Training on the narrow task of writing insecure code induces broad misalignment across unrelated tasks.