Skip to content

Back to the catalog

SecAgg++: A Post-Quantum Privacy Preserving Federated Learning Protocol
eprint.iacr.orgpaper

SecAgg++: A Post-Quantum Privacy Preserving Federated Learning Protocol

SecAgg++ is a masking-based secure aggregation protocol for single-server federated learning that uses post-quantum cryptographic primitives. It keeps the communication and computation efficiency of SecAgg and SecAgg+ and handles client dropout through an efficient recovery procedure. It matters because most masking-based secure aggregation protocols are not post-quantum secure, which leaves federated model updates exposed to future quantum adversaries.

TL;DR Post-quantum successor to SecAgg/SecAgg+ for privacy-preserving federated learning, with efficient client dropout recovery.
Open original ↗
Also seen from another angle🛡️ Security →
Document

What it claims

SecAgg++ is a masking-based secure aggregation protocol for single-server federated learning that swaps in post-quantum primitives. The authors claim it keeps the communication and computation efficiency of SecAgg and SecAgg+, and that it handles client dropout through a recovery procedure. The page's keywords list secret sharing and additive masking. That suggests the SecAgg-style structure: pairwise additive masks, with secret-shared seeds or keys for dropout recovery. The abstract does not spell out that mechanism, so treat it as a reading of the keywords, not a stated design.

How to use it

  • Venue: the paper is peer-reviewed. It was published at AISec 2026, the 19th ACM Workshop on Artificial Intelligence and Security.
  • DOI: 10.1145/3847352.3848087
  • Short URL: https://ia.cr/2026/2497
  • Licence: the ePrint version is licensed CC BY. That covers the paper text. The page mentions no code release and no implementation licence.
  • Authors: all six contact authors have FBK addresses (Fondazione Bruno Kessler).
  • Perez
  • Cavicchioni
  • Cretti
  • Longo
  • Tomasi
  • Ranise
  • Dates: received and approved on ePrint on 2026-10-15. A "See all versions" link exists, but only this entry is shown.

Limits

The fetched page has only the abstract and metadata. It does not establish any of the following, so check the full PDF before relying on the protocol:

  • Primitives: which post-quantum primitives are used. For example, which KEM or key-agreement scheme replaces the Diffie–Hellman step of SecAgg.
  • Efficiency: any cost figures. "Preserves efficiency" comes with no numbers, no baseline measurements against SecAgg or SecAgg+, and no client counts, vector sizes or dropout rates.
  • Threat model: whether security holds against a semi-honest or a malicious server, what collusion thresholds apply, and whether there is a formal security proof.
  • Dropout recovery: what makes the recovery procedure "efficient", and how it compares with SecAgg+'s.
  • Implementation: whether code or a benchmark harness is available.

Scope is stated as single-server only. Multi-server and decentralized settings are not claimed.

Organisations named here

  • AISec
  • SecAgg++

Technologies named here

  • SecAgg
  • SecAgg+
  • SecAgg++
  • Federated Learning
  • Post-Quantum Cryptography
  • Secret Sharing
  • Secure Aggregation

Techniques named here

  • Secure aggregation
  • SecAgg++

Related in research