Mitigating the Privacy Risks of AI through Privacy-Enhancing Technologies
Mitigating the Privacy Risks of AI through Privacy-Enhancing Technologies

Summary/Abstract: The development and operation of an AI solution generally requires large amounts of data. This may involve processing of personal data, which implies privacy risks for the data subjects and the obligation to comply with data protection rules for data controllers. Privacy-enhancing technologies(PETs) can help enhance data collection and mitigate privacy risks posed by the development of AI solutions. In this context, this thesis proposes to present a set of emerging technologies that address privacy risks characteristic to machine learning models and enable privacy-preserving machine learning. The essay will highlight three state-of-the-art PET solutions: homomorphic encryption, secure multi-party computation, and differential privacy.

• Details
• Contents