OpenMined builds open source software which enables external researchers to evaluate proprietary AI systems

At the present moment, many powerful AI products and services remain unchecked by external parties. At the same time, the few external parties who do get to check these properties are often given too much access, exposing AI owners to  numerous risks. This works against the interests of all parties involved: auditors, AI developers, and the billions of people who rely upon increasingly powerful AI products and services in their daily lives. Fixing this external access problem could improve AI products and services, strengthen public trust in AI, mitigate key risks to society,AI developers, and users during the audit process. Addressing this problem has imminent practical relevance given the large and growing impact of AI on our daily lives.

Since 2020, OpenMined has been working on this problem by developing open source, privacy preserving software which enables external researchers to evaluate proprietary AI models. Using this software, external researchers can answer specific questions about a closed-source AI system and its interactions with users without revealing information outside of the answers to those specific questions. During the process of developing, testing, and deploying this software, we have worked closely with a number of leading AI companies, governments, and non-profits, such as those listed below.

Find out more about AI Auditing from OpenMined

Requesting confirms you agree with our privacy policy.