SYNTHESIS NOTE
Topics›MechInterp›this note

Can training data edits reliably override what models already believe?

Synthetic document finetuning appears to add new associations predictably but struggles to revise existing ones. This matters because unpredictable effects cannot be controlled, even if made stronger.

Synthesis note · 2026-09-24 · sourced from MechInterp

The abstract makes a general claim beyond the reward-hacking experiment: "We show that SDF can predictably steer downstream generalization when inserting new associations, but struggles and has unpredictable effects when overriding existing associations, such as that between reward hacking and misalignment that produces EM." The claim has two halves. Insertion works, and override does not. The link from reward hacking to misalignment is the paper's example of an existing association, which puts it in the base model before any RL stage.

That framing turns a specific failure (Can advance document training prevent reward hacking misalignment?) into a statement about what training-data interventions can do. Synthetic documents that say something new to the model add to what it has. Documents that contradict what the model already holds have to contest it, and the paper finds the result of that contest unpredictable, not merely weak. The unpredictability matters more than the weakness, because a weak effect could be scaled up and an unpredictable one cannot be planned around. The abstract's closing line says the same thing: SDF can "steer its generalization from later training in unintended ways."

How it fits the vault (vault reading, not the paper's). Two existing accounts of emergent misalignment already say the base model holds what EM draws on. Does learning to reward hack cause emergent misalignment in agents? carries the account that EM strengthens an existing misaligned persona, and Does representational distance predict where misalignment emerges? predicts EM from the base model's own geometry. This paper adds that documents telling the model otherwise do not remove the association. It does not itself invoke persona or distance, and How is emergent misalignment different from persona changes? shows the vault's EM accounts are contested, so the fit is suggestive only. A parallel on the persistence side is How much poisoned training data survives safety alignment?, where what is learned early survives later intervention.

What the excerpt does not give. Any experiment for the insertion half. The abstract asserts it and the excerpt describes only the reward-hacking case. It also gives no account of how "existing" and "new" associations were told apart, nor the scales, which the paper says bound the claim.

Inquiring lines that read this note 10

This note is a source for these research framings, grouped by the broader line of inquiry each explores. Scan the bold lines of inquiry; follow any specific question forward.

How can we prevent synthetic content from corrupting knowledge corpora? Can AI systems safely improve themselves recursively? How does training on self-generated data affect model capabilities? How does model scale change which features and patterns models learn? Does prompt optimization inject genuinely new knowledge into trained models? How do language models integrate parametric and contextual knowledge? What mechanisms cause models to develop misaligned objectives during training? Why are hallucinations robust to training-based intervention strategies? Do pretraining and finetuning change model capabilities or only output behavior? Can prompt engineering eliminate systematic biases or merely disguise them?

Related concepts in this collection 5

This note in its neighbourhood — explore the map, then jump to a related concept in the list below.

Concept map
14 direct connections · 66 in 2-hop network ·medium cluster Open in graph ↗

Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph

your link semantically near linked from elsewhere

Related papers in this collection 8

Papers most semantically related to this note, ranked by cosine similarity in the embedding space.

Original note title

synthetic document finetuning steers downstream generalization predictably when inserting new associations and unpredictably when overriding existing ones such as reward hacking leading to misalignment