End-to-end encrypted (E2EE) messaging is a modern success story for user privacy. Even if the servers of iMessage, WhatsApp, or Signal were breached, the content exchanged over those platforms would not be available to attackers. This threat model, though strong, does not account for harms that may arise from the content itself. Hate, harassment, and misinformation are several examples of such harms. Platforms protect users from abuse via content moderation. Yet such mitigations often require visibility of the content, which is at odds with end-to-end encryption. This tension has led to calls for “private” scanning of E2EE content, which have been sharply criticized by privacy advocates. This dissertation presents novel techniques for balancing user safety with the strong privacy guarantees of end-to-end encryption. In particular, I investigate how providing enhanced tools for community moderation and content reporting can protect users without compromising on their privacy or agency. First, I introduce a novel framework for private hierarchical governance, which enables rich governance features for encrypted messaging groups while allowing for platform-driven moderation. I then go on to propose a new cryptographic primitive called transcript franking, which extends prior work on reporting encrypted messages to handle sequences of messages with integrity over their causal ordering. Finally, I propose shared encrypted state franking, which enables the maintenance and reporting of shared state in E2EE platforms.