Cloudflare Research logo

A Symbolic Analysis of Privacy for TLS 1.3 with Encrypted Client Hello

Karthikeyan Bhargavan, Vincent Cheval, Christopher A. Wood
ACM Conference on Computer and Communications Security (CCS) 2022, November 7-11, Los Angeles, U.S.A. 2022.


TLS 1.3, the newest version of the Transport Layer Security (TLS) protocol, provides strong authentication and confidentiality guarantees that have been comprehensively analyzed in a variety of formal models. However, despite its controversial use of handshake meta-data encryption,the privacy guarantees of TLS 1.3 remain weak and poorly understood. For example, the protocol reveals the identity of the target server to network attackers, allowing the passive surveillance and active censorship of TLS connections. To close this gap, the IETF TLS working group is standardizing a new privacy extension called Encrypted Client Hello (ECH, previously called ESNI), but the absence of a formal privacy model makes it hard to verify that this extension works. Indeed, several early drafts of ECH were found to be vulnerable to active network attacks.

In this paper, we present the first mechanized formal analysis of privacy properties for the TLS 1.3 handshake. We study all standard modes of TLS 1.3, with and without ECH, using the symbolic protocol analyzer ProVerif. We discuss attacks on ECH, some found during the course of this study, and show how they are accounted for in the latest version. Our analysis has helped guide the standardization process for ECH and we provide concrete privacy recommendations for TLS implementors. We also contribute the most comprehensive model of TLS 1.3 to date, which can be used by designers experimenting with new extensions to the protocol. Ours is one of the largest privacy proofs attempted using an automated verification tool and may be of general interest to protocol analysts.