# ATLAS: Constitution-Conditioned Latent Geometry and Redistribution Across Language Models and Neural Perturbation Data

arXiv: https://arxiv.org/abs/2604.17663 · HTML: https://mlsanity.com/paper/2604.17663/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 35 references that are in the map.
- First posted on arXiv 19 Apr 2026.

## Builds on (most cited first, 22 of 22)

- [Similarity of Neural Network Representations Revisited](https://mlsanity.com/paper/1905.00414/) (2019)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [Refusal in Language Models Is Mediated by a Single Direction](https://mlsanity.com/paper/2406.11717/) (2024)
- [Steering Language Models With Activation Engineering](https://mlsanity.com/paper/2308.10248/) (2023)
- [The Platonic Representation Hypothesis](https://mlsanity.com/paper/2405.07987/) (2024)
- [Function Vectors in Large Language Models](https://mlsanity.com/paper/2310.15213/) (2023)
- [AI Control: Improving Safety Despite Intentional Subversion](https://mlsanity.com/paper/2312.06942/) (2023)
- [AI Sandbagging: Language Models can Strategically Underperform on Evaluations](https://mlsanity.com/paper/2406.07358/) (2024)
- [Collective Constitutional AI: Aligning a Language Model with Public Input](https://mlsanity.com/paper/2406.07814/) (2024)
- [Model Organisms for Emergent Misalignment](https://mlsanity.com/paper/2506.11613/) (2025)
- [Thought Crime: Backdoors and Emergent Misalignment in Reasoning Models](https://mlsanity.com/paper/2506.13206/) (2025)
- [Harnessing the Universal Geometry of Embeddings](https://mlsanity.com/paper/2505.12540/) (2025)
- [Specific versus General Principles for Constitutional AI](https://mlsanity.com/paper/2310.13798/) (2023)
- [A sketch of an AI control safety case](https://mlsanity.com/paper/2501.17315/) (2025)
- [The Hawthorne Effect in Reasoning Models: Evaluating and Steering Test Awareness](https://mlsanity.com/paper/2505.14617/) (2025)
- [Transferring Linear Features Across Language Models With Model Stitching](https://mlsanity.com/paper/2506.06609/) (2025)
- [How to evaluate control measures for LLM agents? A trajectory from today to superintelligence](https://mlsanity.com/paper/2504.05259/) (2025)
- [Steering Vector Fields for Context-Aware Inference-Time Control in Large Language Models](https://mlsanity.com/paper/2602.01654/) (2026)
- [On the Non-Identifiability of Steering Vectors in Large Language Models](https://mlsanity.com/paper/2602.06801/) (2026)
- [ABC Align: Large Language Model Alignment for Safety & Accuracy](https://mlsanity.com/paper/2408.00307/) (2024)
- [Where to Steer: Input-Dependent Layer Selection for Steering Improves LLM Alignment](https://mlsanity.com/paper/2604.03867/) (2026)
- [ENIGMA: The Geometry of Reasoning and Alignment in Large-Language Models](https://mlsanity.com/paper/2510.11278/) (2025)

## Built on by (most cited first, 1 of 1)

- [Order Is Not Control: Driven-Dissipative Response Laws Across Artificial and Biological Systems](https://mlsanity.com/paper/2606.12923/) (2026)

## Abstract (arXiv)

Constitution-conditioned post-training can be analysed as a structured perturbation of a model's learned representational geometry. We introduce ATLAS, a geometry-first program that traces constitution-induced hidden-state structure across charts, models, and substrates. Instead of treating the relevant unit as a single behaviour, neuron, vector, or patch, ATLAS tests a local chart whose tangent structure, occupancy distribution, and behavioural coupling can be measured under system change. On Gemma, the anchored source-local chart captures 310 / 320 reviewed source rows and all 84 / 84 reviewed score-flip rows, but compact exact-patch sufficiency does not close, so the exportable unit is the broader source-defined family. Freezing that family, we re-identify a target-local realisation in an unadapted Phi model, where the fully adjudicated confirmatory contrast separates with AUC 0.984 and mean gap 5.50. In held-out ALM8 mouse frontal-cortex perturbation data, the same source-defined family receives support across 5/5 folds, with mean held-out AUC 0.72 and mean fold gap 4.50. A multiple-choice analysis provides the main boundary: nearby target-local signals can appear without source-faithful closure. The resulting correspondence is not coordinate identity, site identity, or a target-side mediation theorem. It is geometric recurrence under redistribution: written constitutions can induce recoverable latent geometry whose organisation remains detectable across model and substrate changes while its local coordinates, occupancy, and behavioural expression shift.