A team of cryptographic researchers has published a new study revealing a significant privacy vulnerability affecting widely used multi-party computation (MPC) frameworks. The research paper, titled “Off-Key: Inconsistent MAC Keys Compromise Privacy in SPDZ-Family MPC,” was authored by Marcel Keller of CSIRO Technology, Sela Navot of Artemis Software Works, and Nirvan Tyagi of the University/Washington. Released on the Cryptology ePrint Archive as Paper 2026/2533, the study examines how subtle inconsistencies in message authentication code keys can lead to severe privacy breaches in protocols designed for maliciously secure computations.
Multi-party computation forms a cornerstone of modern cryptographic engineering, allowing multiple distrustful parties to jointly compute a function over their private inputs without revealing those inputs to one another. Within this broad domain, the SPDZ family of protocols stands out as a prominent paradigm for maliciously secure computation—settings where participants may actively attempt to cheat, deviate from instructions, or disrupt the protocol execution. To protect sensitive secret-shared values from tampering, SPDZ-family protocols rely heavily on message authentication codes, commonly referred to as MACs. These MACs are generated under a secret-shared global key distributed among the participants, ensuring that any unauthorized manipulation of the data during computation will be detected during validation checks.
The foundational security requirement of this mechanism is strict uniformity: all values and preprocessing material utilized together within an execution must be authenticated under the exact same global key. However, the new research demonstrates that when this requirement is violated, malicious participants can exploit the resulting state inconsistencies to compromise the privacy of shared secrets. The authors introduce “Off-Key,” a novel privacy attack that targets these vulnerabilities across various execution environments.
Depending on the specific MAC check protocol implemented in a given software version, a corrupted party leveraging the Off-Key attack can extract the value of a shared secret from a single authentication check. Alternatively, the attacker can test a guess about a secret through selective failure. By modifying a MAC share and observing whether a masked verification check succeeds or fails, the adversary can systematically deduce hidden information. While the authors note that this attack does not strictly break the fundamental theoretical security guarantees established by the underlying cryptographic protocols, it demonstrates a critical gap between theoretical design and practical implementation.
To assess the real-world impact of their findings, the research team evaluated popular open-source multi-party computation libraries and frameworks, specifically targeting MP-SPDZ, SCALE-MAMBA, and FANNG-MPC. Their investigation revealed that evaluated versions of these prominent frameworks inadvertently permit a corrupted party to utilize different keys across operational boundaries, such as across separate batches, independent threads, or distinct process runs. These structural allowances create the exact window of vulnerability required for the Off-Key attack to succeed in practice.
Addressing the issue requires a careful re-evaluation of how state and key consistency are managed during the preprocessing phases of multi-party computation. To assist software developers and system architects in safeguarding their implementations against this class of attacks, the researchers identified the precise internal state responsible for enforcing key consistency in representative preprocessing protocols. Furthermore, they demonstrated how properly preserving and reusing this state throughout the execution lifecycle entirely prevents the vulnerability from manifesting.

