History
Probes on sparse-autoencoder latents generalize better than raw-activation probes when trained on synthetic data and tested on real deployment data.
Claim entered the graph
Nothing has happened to this claim yet beyond its creation: no assessments, contributions, or decisions are on record.