# Preprint demonstrates encrypted model capabilities that resist audit methods

_Published Tuesday, October 13, 2026 at 3:19 AM EDT · Science, AI · Latest · Tier 2 — Notable_

Researchers describe a way to conceal arbitrary computations inside AI model weights in a new preprint, challenging the assumption that capability audits can elicit every hidden behavior. Their construction embeds encrypted neural networks that decrypt and execute only when a model receives a secret trigger.

The researchers implemented proof-of-concept attacks in PyTorch and tested them on a small chemical reaction prediction task. The hidden capabilities resisted fine-tuning, weight noising, gradient-based trigger searches and latent adversarial training. Their cryptographic guarantees depend on assumptions, and the chemical task serves as a proxy for capabilities relevant to chemical, biological, radiological and nuclear risks.

## Sources

- [arxiv.org](https://arxiv.org/abs/2610.13928)

---
Canonical: https://techandbusiness.org/newswire/r2CAco4IGO5vUsWgq6DDtS
Published: 2026-10-13T07:19:12.435Z
Story chronology: 2026-10-13T02:07:08.000Z
Retrieved: 2026-10-13T09:40:00.583Z
Publisher: Tech & Business (techandbusiness.org)