# Goodfire finds retrained probes recover signals that break in language models

> Original title: We investigated this with both theory and experiments. Probes can break, but predictably: the concept usually moves rather than disappear...

- Source: Goodfire
- Published: 2026-11-05T17:21:08.000Z
- Bad Influence: https://badinfluence.news/items/xatozqvld7vib2y683jptpqb5
- Original: https://x.com/GoodfireAI/status/2118393189845667911

## Summary

Goodfire reports that probes for language models can break, but usually the concept moves rather than disappears. In its language model experiments, retraining a probe was enough to recover the signal.

## Post

We investigated this with both theory and experiments. Probes can break, but predictably: the concept usually moves rather than disappears. In our language model experiments, retraining a probe was enough to recover the signal.

[Video](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Famplify_video_thumb%2F2118393177828970496%2Fimg%2FRrWC3mA2PPtPPFNy.jpg&mode=thumb&exp=1794096000&sig=b22a34a25d162d07)