cs.LGNov 18, 2026

Stealing Exact Training Data from Other Clients during Federated Learning

Authors: Omkar Shrikanth, Soham Dambalkar, Parag Jyoti Pal, Advik Raj Basani, Arnab K. Paul

Organizations: Birla Institute of Technology and Science, Goa · MATS Research.

Abstract

Malicious adversaries can employ gradient inversion attacks to reverse-engineer private data using the model updates of a round in Federated Learning. While there have been developments in gradient inversion for text, exact text recovery by a malicious client has so far remained infeasible, as FedAvg entangles the gradient contributions of many clients and obscures the signal of any individual sentence. We introduce a novel attack pipeline that unwinds this entanglement progressively through an iterative loop, in which the gradient contributions of sentences recovered exactly in one round are subtracted from the aggregate, exposing the remaining sentences for recovery in the next iteration. We show, for the first time, exact sentence recovery, with 100% precision, and on aggregates containing up to 4×4\times more tokens than the model's hidden dimension, we recover up to 80.2% of client sentences exactly, a regime where existing attacks recover none exactly.

Figures & tables

Appendix figures & tables15 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Cascading Gradient Inversion via LT-Code Inspired Peeling in Federated Learning

    Sep 9, 2026Saeed Shariati, Mohsen Alambardar MeybodiGradient Inversion AttacksFederated Learning

  2. No More Guessing: a Verifiable Gradient Inversion Attack in Federated Learning

    Apr 16, 2026Francesco Diana, Chuan Xu, André Nusser +1Gradient Inversion AttacksAdversarial Attacks

  3. Gradient Inversion through Linear and Affine Subspace Constraints

    Nov 10, 2026Konstantinos Bouros, Athanasios P. LiavasGradient Inversion AttacksTraining Data Reconstruction Attacks