{"ID":6138744,"CreatedAt":"2026-07-09T01:07:32.349475501Z","UpdatedAt":"2026-07-10T15:16:26.541449957Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.06643","arxiv_id":"2607.06643","title":"The Power of Backdoor Absorption in Community Training","abstract":"Backdoor attacks severely threaten large-scale AI models. When model owners delegate training to external compute providers within a decentralized training paradigm, adversaries can craft stealthy, low-frequency triggers to inject malicious behavior while evading standard audits. Traditionally, detecting these attacks requires a full re-computation of the training steps--a prohibitive overhead that directly contradicts the owner's resource constraints. To address this, we investigate the resilience of continuous optimization dynamics under Byzantine perturbations, where adversaries are forced to compete against a continuous influx of honest updates. Under a threat model where an adversary compromises f out of n total trainers, we quantify the minimum auditing overhead required by the model owner to probabilistically bound the attack success rate. We formalize this injection-absorption dynamic as a Discrete-Time Markov Chain (DTMC). Using this framework, we prove that the success probability of any bounded adversary asymptotically collapses to zero under a defense strategy combining natural absorption, a randomized scheduler, and lazy verification oracle. Empirical results demonstrate significant backdoor suppression with zero utility degradation even when invoking the verification oracle on merely 10% of the total training steps. This approach yields a provably sound and computationally efficient defense for safety-critical AI.","short_abstract":"Backdoor attacks severely threaten large-scale AI models. When model owners delegate training to external compute providers within a decentralized training paradigm, adversaries can craft stealthy, low-frequency triggers to inject malicious behavior while evading standard audits. Traditionally, detecting these attacks...","url_abs":"https://arxiv.org/abs/2607.06643","url_pdf":"https://arxiv.org/pdf/2607.06643v1","authors":"[\"Issam Seddik\",\"Sami Souihi\",\"Mohamed Tamaazousti\",\"Sara Tucci Piergiovanni\"]","published":"2026-07-07T15:19:20Z","proceeding":"cs.CR","tasks":"[\"cs.CR\",\"cs.LG\"]","methods":"[]","has_code":false}