Current centralized systems bring numerous drawbacks, both in terms of cost and security, especially when it comes to privacy issues. As an alternative, decentralized can overcome some of these problems by lowering the trust needed to run the system. Blockchain is the main technology used to build a decentralized system, as it provides a distributed architecture secured by publicly-verifiable data immutability. Nonetheless, blockchain transparency can jeopardize users' privacy as transactions and their metadata may contain sensitive information. This led to the development of privacy-preserving blockchains -- such as ZCash and Monero -- which enable users to make transactions anonymously. Yet, these frameworks implement privacy protections to the detriment of accountability, which might be problematic for some applications such as decentralized learning for healthcare and finance.  This PhD proposal aims at investigating how to provide privacy guarantees for the majority of the blockchain users while still being able to make dishonest users accountable for their actions on the network.