Buy Aurelius (SN37) in France

Aurelius, identified as Subnet 37 on the Bittensor network, is a decentralized crypto and Web3 project dedicated to the field of AI safety and alignment. The primary mission of the project is to ensure that artificial intelligence systems act in accordance with human values and intentions by creating a transparent and verifiable process for model evaluation. The project operates as an incentive-driven protocol where participants collaborate to identify and document failures in large language models. The architecture of Aurelius is built around three distinct roles: miners, validators, and the Tribunate. Miners act as adversarial prompters who perform red-teaming on models to surface issues like bias, toxicity, and hallucinations. Their work generates synthetic alignment data that highlights where models may be failing or behaving dishonestly. Validators serve as independent auditors who verify the evidence provided by miners and score the results based on established safety rubrics. The Tribunate functions as the governance layer, maintaining a living constitution or a set of rules that defines what constitutes safe and aligned AI behavior. The core product delivered by the Aurelius subnet is a continuous stream of high-quality, structured datasets. These datasets are intended for use by developers and researchers to fine-tune AI models, moving beyond simple knowledge training to incorporate a layer of wisdom and ethical reasoning. By decentralizing the alignment process, Aurelius aims to overcome the limitations and potential biases of centralized safety testing performed by private labs. The project is developed by Aurelius Labs and is led by its founder, Austin McCaffrey. It transitioned into its current form on the Bittensor network in late 2025, taking over the Subnet 37 slot to focus specifically on the growing need for verifiable AI safety data. Through its peer-to-peer ecosystem, Aurelius seeks to establish a decentralized watchdog layer for the next generation of artificial intelligence, providing the tools necessary to audit and refine frontier models in an open and collaborative environment. Learn more