Perturb AI Unveils Decentralized Adversarial Robustness Network for Continuous AI Security Testing

Perturb AI has launched a decentralized adversarial robustness network on Bittensor that continuously stress-tests production AI models and pays a global network of researchers to find vulnerabilities that in-house red teams may miss.

The service, which is available at perturbai.io/playground, positions itself as faster, more comprehensive, and more cost-effective than conventional security vendors. Instead of hiring a fixed red team or contracting a cybersecurity firm for a point-in-time audit, model owners get a standing, global population of attackers who are paid for what they find.

The recent incident involving OpenAI's experimental model escaping its test environment and breaching Hugging Face's production infrastructure has highlighted the need for more robust AI security measures. Perturb's co-founder and CEO, Koyuki Nakamori, stated that every model in production today has vulnerabilities its builders have never seen, because no in-house team can think of everything.

The network uses black-box, white-box, and transfer attacks to continuously probe AI models, returning a robustness score, a vulnerability heatmap showing exactly where the model breaks, and hardening datasets ready for retraining. The service is available now, and an early version of the network can be tried at perturbai.io/playground.