# VFUSE: Virulent Feature Understanding with Sparse autoEncoders

arXiv: https://arxiv.org/abs/2606.10080 · HTML: https://mlsanity.com/paper/2606.10080/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Interpreting biomedical and science models, one of 13 directions in Mechanistic interpretability.
- Builds on 10 of its 28 references that are in the map.
- First posted on arXiv 8 Jun 2026.

## Placed in

- Mechanistic interpretability / [Interpreting biomedical and science models](https://mlsanity.com/mechinterp/interpreting-biomedical-and-science-models/)

## Builds on (most cited first, 10 of 10)

- [Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps](https://mlsanity.com/paper/1312.6034/) (2013)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Scaling and evaluating sparse autoencoders](https://mlsanity.com/paper/2406.04093/) (2024)
- [Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet](https://mlsanity.com/paper/2605.29358/) (2026)
- [Learning Multi-Level Features with Matryoshka Sparse Autoencoders](https://mlsanity.com/paper/2503.17547/) (2025)
- [BatchTopK Sparse Autoencoders](https://mlsanity.com/paper/2412.06410/) (2024)
- [InterPLM: Discovering Interpretable Features in Protein Language Models via Sparse Autoencoders](https://mlsanity.com/paper/2412.12101/) (2024)
- [Prisma: An Open Source Toolkit for Mechanistic Interpretability in Vision and Video](https://mlsanity.com/paper/2504.19475/) (2025)
- [Towards Interpretable Protein Structure Prediction with Sparse Autoencoders](https://mlsanity.com/paper/2503.08764/) (2025)
- [SafeProtein: Red-Teaming Framework and Benchmark for Protein Foundation Models](https://mlsanity.com/paper/2509.03487/) (2025)

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

- [Artificial intelligence and biosecurity: capabilities, threat pathways, and defense-in-depth governance](https://mlsanity.com/paper/2609.16213/) (2026)

## Abstract (arXiv)

Generative models have shown remarkable progress in a variety of domains such as protein design, but such power enables the opaque generation of hazardous proteins. In this work, we introduce VFUSE (Virulent Feature Understanding with Sparse autoEncoders), a mechanistic interpretability approach that trains SAEs on diffusion-transformer activations to audit protein models for hazard-aware features. We apply VFUSE to RoseTTAFold3 and RFDiffusion3, popular open-weight models for protein folding and synthesis. We find that for certain blocks, linear probes detect hazardous designs significantly better when fit in the SAE latent space over the original model's representations: improving interpretability without sacrificing model performance. Furthermore, we identify monosemantic features from the SAE that fire only on hazardous designs at up to AUROC 0.84 (q < 10-13). To our knowledge this is the first SAE trained on an all-atom diffusion model and the first feature-level virulence audit of a protein design model, paving the way towards safe and interpretable protein design.