- Rating
- 3(+1)
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper investigates refusal backdoor propagation in decentralized LLM fine-tuning (D-PSGD with LoRA), showing that a single malicious node can reach up to 78% ASR on honest neighbors. The authors propose CHORUS, a two-stage behavioral screening protocol where co-receivers collaborate across 2-hop neighborhoods: Stage 1a probes refusal rates on arbitrary triggers via majority vote, Stage 1b harvests recurring strings via beam search to test log-probability "movability," and Stage 2 manages neighbor trust states. Experiments on Llama-2-7B-Chat and Qwen3-8B across Alpaca and Dolly-15k on a 16-node 3-regular graph show CHORUS reduces neighbor ASR to and outperforms decentralized AlignIns.
Critical strengths
- Complementary Behavioral Screening for Decentralized Topologies: The paper clearly demonstrates why weight-space outlier detection fails in small, non-IID decentralized neighborhoods (>34% FPR floor). Combining Stage 1a (effective in early rounds before backdoor saturation) and Stage 1b (effective once target strings are memorized) is well-motivated and achieves 100% TPR on the tested setups with only 404 bytes of communication overhead per round.
- Clear Exposition and Diagnostic Analysis: The manuscript is easy to follow and provides insightful diagnostics on how backdoor propagation evolves across graph hops and why Stage 1a sensitivity decays in later rounds (Appendix E.4).
Critical weaknesses
- Insufficient Experimental Scope on Backdoor Types and Adaptive Attacks: The empirical evaluation is restricted to a single verbatim refusal string ("I'm sorry, I cannot help with that request.") and single-word triggers ("BadMagic"). Both stages rely heavily on this narrow setup: Stage 1a matches an 18-phrase refusal list, while Stage 1b requires an exact target string of words to recur across beam search and show near-zero movability (). Furthermore, the adaptive attacker in Appendix E.3 only targets Stage 1a; there are no experiments evaluating adaptive attacks against Stage 1b (e.g., paraphrased/multi-target refusals) or non-refusal backdoors (e.g., sentiment steering, bias injection, or short targets).
- Lack of Topology, Scalability, and Baseline Experiments: All experiments are fixed to a single, small, symmetric 16-node 3-regular circulant graph. There is no evaluation on larger networks, irregular/scale-free topologies, dynamic graphs, or varying node degrees (e.g., , where strict majority voting breaks down under a single colluding co-receiver). Moreover, only a single defense baseline is evaluated in the main text, while the multi-attacker results ( in Appendix E.1) rely on offline log replays and a modified Stage 1a rule rather than full end-to-end runs due to CHORUS's high wall-clock compute overhead.
Questions for authors
- Paraphrased / Multi-Target Backdoors and Stage 1b Adaptive Evasion: How does CHORUS perform when an attacker uses diverse/paraphrased refusal targets (or non-refusal backdoor objectives) rather than a single verbatim target string? Can an adaptive attacker evade both stages simultaneously by combining the Stage 1a evasion loss (Appendix E.3) with multi-target poisoning that prevents exact string recurrence in HARVESTSTRINGS?
- Graph Topologies and Scalability: Can the authors provide end-to-end experiments on non-circulant topologies (e.g., Erdős–Rényi or uneven node degrees) and larger graphs (), as well as full end-to-end training runs (rather than log replays) for multiple colluding attackers ()?
- False Positives on Domain-Specific Boilerplate: Under non-IID partitions where an honest node trains on data with recurring boilerplate (e.g., medical/legal disclaimers unseen by its neighbors), does Stage 1b falsely flag the honest node due to low movability on those memorized phrases?
3 comments
Thanks for the reply. I have raised the score to support this paper acceptance.
Official Review of Submission58039
thank you for the answers. I raise my score.
We thank the reviewer for their detailed review, and adress their concerns below.
We thank the reviewer for this question. CHORUS needs neither the attacker's trigger nor its target string. It detects the change that a refusal backdoor leaves in the model. ICLScan shows that a backdoored model extends its trigger-to-refusal shortcut to triggers it was never trained on (their Figure 3 and Table 2). Stage 1a builds on this insight and only has to recognize that a response is a refusal, whatever its wording. In our revision (Appendix G), we replace the substring heuristic used for this decision with a small encoder-based refusal classifier with a reported F1 score of 0.9874, which does not depend on a fixed list of phrases. In our main setting (using Llama-2-7B-Chat on the Alpaca dataset, n=16, 3-regular graph, one attacker), CHORUS with the classifier reaches a TPR of 100% at an FPR of 0.89%, and the classifier adds less than 1% to the total runtime.
Stage 1b, in contrast, looks for a string that the sender has memorized. An attacker that spreads its target over many paraphrases is an adaptive attack on this stage, and CHORUS then relies on Stage 1a, since an update is rejected if either test flags it. We consider such adaptive attacks, together with non-refusal objectives, an interesting direction for future work.
Topology. We thank the reviewer for this suggestion. We ran the Erdos-Renyi graph as suggested, with n=16 nodes and report the results in the table below. In order to be fair to the baseline AlignIns, we made sure each of the three neighbors of the attacker also gets 3 neighbors. This is not needed for CHORUS but required for AlignIns, which cannot reject anything if only 1 or 2 neighbors. As shown below, CHORUS matches the Oracle perfectly for this run, with an perfect TPR and FPR.
Larger graphs. We remark that CHORUS is a local method: a node only communicates with its direct neighbors (exchanging model updates) and with its distance-2 neighbors (for the collaborative steps). As such, increasing the number of nodes in the network, assuming node degrees do not change, does not impact the computational cost or the effectiveness of the defense. Besides, while increasing the degree does increase the computational cost, updates are still screened independently. We refer to the additional experiment above, where degree ranged from 1 to 6.
Multiple colluding attackers. We would like to clarify that adversaries are by default colluding: they share the same trigger and target refusal response. To address the reviewer's concern, we re-ran the m=2 run on our defaut topogy from scratch (without replaying the logs). CHORUS achieves ASR 1.11 with 100% TPR (144/144) and 0.23% FPR (2/864), and loss 1.243. This reproduces the paper's row almost identically. We'll add the results for m=3 when it finishes.
As in our answer to Q1/W1, we emphasize that CHORUS does not look for a trigger or for a particular string. It measures how the behavior of an adapter changes when a trigger is introduced into the prompt. In Table 9, we probe the same adapters with triggers of varying levels of surprise, from a nonsense token (
ctfqxy) to a word as common as the. With each of them, honest adapters score below the detection threshold. Domain-specific boilerplate lies within this range.