Summary
In this chapter, we covered privacy attacks aiming to steal data by means of reconstructing training data with model inversion attacks or inferring global or instance data with attribute and membership inference attacks.
We discussed several mitigations. An underlying theme of these defenses has been the need to prevent data privacy. The following chapter will explore in detail the field of privacy-preserving AI, which includes a variety of techniques that help us minimize sensitive data exposure and protect privacy from the ground up.
