# The Truth Was Never Gone: Perfect Aliasing in Compliant-Context Truth Probes

arXiv: https://arxiv.org/abs/2609.10739 · HTML: https://mlsanity.com/paper/2609.10739/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Probing deception, bias and beliefs, one of 13 directions in Mechanistic interpretability.
- Builds on 13 of its 13 references that are in the map.
- First posted on arXiv 9 Sep 2026.

## Placed in

- Mechanistic interpretability / [Probing deception, bias and beliefs](https://mlsanity.com/mechinterp/probing-deception-bias-and-beliefs/)

## Builds on (most cited first, 13 of 13)

- [The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets](https://mlsanity.com/paper/2310.06824/) (2023)
- [Natural Emergent Misalignment from Reward Hacking in Production RL](https://mlsanity.com/paper/2511.18397/) (2025)
- [Detecting Strategic Deception Using Linear Probes](https://mlsanity.com/paper/2502.03407/) (2025)
- [Challenges with unsupervised LLM knowledge discovery](https://mlsanity.com/paper/2312.10029/) (2023)
- [RL-Obfuscation: Can Language Models Learn to Evade Latent-Space Monitors?](https://mlsanity.com/paper/2506.14261/) (2025)
- [Strategic Dishonesty Can Undermine AI Safety Evaluations of Frontier LLMs](https://mlsanity.com/paper/2509.18058/) (2025)
- [The Obfuscation Atlas: Mapping Where Honesty Emerges in RLVR with Deception Probes](https://mlsanity.com/paper/2602.15515/) (2026)
- [The Impact of Off-Policy Training Data on Probe Generalisation](https://mlsanity.com/paper/2511.17408/) (2025)
- [One Probe Won't Catch Them All: Towards Targeted Deception Detection](https://mlsanity.com/paper/2602.01425/) (2026)
- [Pressure-Testing Deception Probes in LLMs: Scaling, Robustness, and the Geometry of Deceptive Representations](https://mlsanity.com/paper/2605.27958/) (2026)
- [How Context Shapes Truth: Geometric Transformations of Statement-level Truth Representations in LLMs](https://mlsanity.com/paper/2601.06599/) (2026)
- [Rift: A Conflict Signature for Deception in Language Models](https://mlsanity.com/paper/2606.17229/) (2026)
- [Diagnosing Correctness Probes under Self-Judgement Confounding](https://mlsanity.com/paper/2607.16799/) (2026)

## Abstract (arXiv)

Linear probes that decode the truth from a language model's activations have been proposed as deception monitors, and a truth probe scoring below chance on a model trained to deceive is naturally read as evidence that the model hid or stopped representing the truth. We show that a fitting-label ambiguity can produce the same readout. In a controlled game, a model should report a secret bit to an ally and its complement to a rival. On compliant (ally) contexts the true bit and the answer the task prescribes are identical labels, so a probe fitted there cannot tell which of the two it measures. We call this complete agreement perfect aliasing. The labels are complements on rival contexts, so one ally-fitted probe scored against each has rival AUROCs that sum to one. Mixed fitting, on ally and rival contexts together, makes the two labels differ; randomized output codebooks also decouple the prescribed answer from the output letter. For a reward-trained Gemma-2-9B policy that answers falsely on every evaluated rival trial, the ally-fitted probe scores 0.006 ± 0.005 AUROC at the final layer (mean ± sample SD over three RL training seeds), while mixed-fit probes score 1.000 on the same held-out activations. Mixed fitting also uses more examples and labelled rival data, so this shows the true bit remains linearly recoverable, not that separating the labels alone explains the gain. In instructed Llama-3.1-8B, a probe refitted on one prompt variant and one frozen from a reference variant both have held-out ally accuracy 1.000 but rival truth AUROCs of 0.080 and 0.986 on the same trials. Our main experiments use one single-token game that states the bit in the prompt, so the recovered direction may read a retained copy of it; where the model must infer the bit, no tested arm deceives reliably. We study what a probe measures, not whether the model uses that information.