# SNNs harder to steal: study

Digest AI · Research · published 2026-10-28T04:00:00Z

Canonical: https://digestai.news/story/snns-harder-to-steal-study

## Summary

A new benchmarking study reveals that spiking neural networks (SNNs) are more resistant to model stealing attacks compared to conventional artificial neural networks (ANNs). The research, published in arXiv cs.AI, tested four representative model stealing attack methods against SNNs and their ANN counterparts on the MNIST, Fashion-MNIST, and CIFAR-10 datasets. Across all attacks and data sets, SNNs proved more difficult to steal. Surrogates of SNNs achieved 10% to 25% lower agreement with the target model and 15% to 25% lower adversarial transferability than surrogates of ANNs. The study attributes this resistance to the unique computing methods of SNNs, which use sparse, discrete spikes and time-dependent membrane potentials that make it harder for attackers to reconstruct decision boundaries.

## Key points

- SNNs are more resistant to model stealing attacks
- Surrogates of SNNs show lower agreement with target models
- Resistance attributed to unique computing methods

## Why it matters

These findings strengthen the case for neuromorphic architectures in security-sensitive applications, as they demonstrate that SNNs are more resilient to model stealing attacks compared to conventional ANNs.

## Sources

1. [Spiking Neural Networks Are Harder to Steal: A Benchmark Study of Model Stealing Attacks](https://arxiv.org/abs/2610.36298) (arXiv cs.AI, 2026-10-28, primary source)

## Cite

Digest AI, "SNNs harder to steal: study", 28 October 2026, https://digestai.news/story/snns-harder-to-steal-study

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/snns-harder-to-steal-study.json