# AntiPaSTO: Self-Supervised Honesty Steering via Anti-Parallel Representations

arXiv: https://arxiv.org/abs/2601.07473 · HTML: https://mlsanity.com/paper/2601.07473/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 27 of its 41 references that are in the map.
- First posted on arXiv 12 Jan 2026.

## Builds on (most cited first, 27 of 27)

- [Concrete Problems in AI Safety](https://mlsanity.com/paper/1606.06565/) (2016)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Steering Language Models With Activation Engineering](https://mlsanity.com/paper/2308.10248/) (2023)
- [Probing Classifiers: Promises, Shortcomings, and Advances](https://mlsanity.com/paper/2102.12452/) (2021)
- [A Simple and Effective Pruning Approach for Large Language Models](https://mlsanity.com/paper/2306.11695/) (2023)
- [Discovering Latent Knowledge in Language Models Without Supervision](https://mlsanity.com/paper/2212.03827/) (2022)
- [Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervision](https://mlsanity.com/paper/2312.09390/) (2023)
- [AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders](https://mlsanity.com/paper/2501.17148/) (2025)
- [Controlling Text-to-Image Diffusion by Orthogonal Finetuning](https://mlsanity.com/paper/2306.07280/) (2023)
- [Programming Refusal with Conditional Activation Steering](https://mlsanity.com/paper/2409.05907/) (2024)
- [Goal Misgeneralization: Why Correct Specifications Aren't Enough For Correct Goals](https://mlsanity.com/paper/2210.01790/) (2022)
- [Goal Misgeneralization in Deep Reinforcement Learning](https://mlsanity.com/paper/2105.14111/) (2021)
- [Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization](https://mlsanity.com/paper/2406.00045/) (2024)
- [Universal Neurons in GPT2 Language Models](https://mlsanity.com/paper/2401.12181/) (2024)
- [Utility Engineering: Analyzing and Controlling Emergent Value Systems in AIs](https://mlsanity.com/paper/2502.08640/) (2025)
- [Understanding (Un)Reliability of Steering Vectors in Language Models](https://mlsanity.com/paper/2505.22637/) (2025)
- [Exposure Bias versus Self-Recovery: Are Distortions Really Incremental for Autoregressive Text Generation?](https://mlsanity.com/paper/1905.10617/) (2019)
- [HyperSteer: Activation Steering at Scale with Hypernetworks](https://mlsanity.com/paper/2506.03292/) (2025)
- [Adaptive Multi-Subspace Representation Steering for Attribute Alignment in Large Language Models](https://mlsanity.com/paper/2508.10599/) (2025)
- [Steering the CensorShip: Uncovering Representation Vectors for LLM "Thought" Control](https://mlsanity.com/paper/2504.17130/) (2025)
- [RepIt: Steering Language Models with Concept-Specific Refusal Vectors](https://mlsanity.com/paper/2509.13281/) (2025)
- [Rethinking Reward Model Evaluation Through the Lens of Reward Overoptimization](https://mlsanity.com/paper/2505.12763/) (2025)
- [DeLoRA: Decoupling Angles and Strength in Low-rank Adaptation](https://mlsanity.com/paper/2503.18225/) (2025)
- [Distribution-Aligned Decoding for Efficient LLM Task Adaptation](https://mlsanity.com/paper/2509.15888/) (2025)
- [Deep Value Benchmark: Measuring Whether Models Generalize Deep Values or Shallow Preferences](https://mlsanity.com/paper/2511.02109/) (2025)
- [SSVD: Structured SVD for Parameter-Efficient Fine-Tuning and Benchmarking under Domain Shift in ASR](https://mlsanity.com/paper/2509.02830/) (2025)

## Abstract (arXiv)

As models grow more capable, humans cannot reliably verify what they say. Scalable steering requires methods that are internal, self-supervised, and transfer out-of-distribution; existing methods satisfy some but not all three. We introduce AntiPaSTO, which separates representations along an antiparallel axis (+1/-1 produce opposite shifts), with coherence constraints preventing collapse. Training uses only two contrasting words inserted into template sentences, with no preference labels. When we use 800 such synthetic pairs on Gemma-3-1B, AntiPaSTO beats prompting baselines by 6.9x Steering F1 on DailyDilemmas and wins on 5 of 6 tested value axes. We also find preliminary evidence that it maintains bidirectional control where prompting triggers refusal.