Fully homomorphic encryption (FHE) allows computation on encrypted data, but decrypting the result usually requires a single secret key. Threshold FHE distributes that decryption across multiple parties, reducing the risk of a single compromised key. The catch is that many threshold constructions are tightly coupled to a specific FHE scheme and assume synchronous participation.

Tulva, described in a new arXiv preprint, decouples threshold decryption from the underlying FHE scheme. The authors present it as a modular add-on: you take an existing FHE scheme, attach Tulva's asynchronous threshold decryption mechanism, and the combination works without redesigning the base scheme. Because the mechanism is asynchronous, parties can contribute decryption shares at their own pace rather than requiring coordinated rounds.

The main practical benefit is flexibility. Instead of building threshold support into a particular FHE library from the ground up, Tulva claims compatibility with virtually any base FHE scheme. That means researchers and engineers could add threshold capabilities to an existing implementation with less friction, though the paper's abstract does not yet describe the exact overhead or security trade-offs involved.