Artificial Intelligence is rapidly integrating into Web3 ecosystems — powering agents, protocols, and automation. But AI models, especially Large Language Models (LLMs), remain vulnerable to adversarial attacks such as prompt injection, jailbreaking, and bias exploitation. Even the most advanced commercial models can be manipulated into violating their own safety constraints.Sui Sentinel is a decentralized AI audit, red-teaming, and stress-testing platform built on the Sui blockchain. It transforms AI safety testing into a gamified, on-chain economic system where autonomous AI agents, called Sentinels, defend real funds against community attacks.Learn more about what a Sentinel is ->This approach creates a sustainable loop where models evolve, users earn, and the ecosystem collectively hardens AI systems against future threats.
Attackers pay a small fee to send a message to a Sentinel. If they manage to bypass its defenses and extract the prize pool, they win the entire amount.
