{"data":{"id":"3578d925-a4d5-407c-8d2d-638fa7013915","title":"Rented Oscillations: Multi-Tenant Data Center Workload Synchronization Attacks and Mitigation","summary":"This research paper shows that a stealthy adversary renting workstations in multi-tenant data centers can run deliberately unoptimized LLM training jobs whose consumption synchronizes into large periodic oscillations, while each job looks benign alone. Simulations and PHIL experiments on a surrogate model of the Saudi Arabian transmission network indicate that MW-scale oscillations saturate generator control limiters and excite a lightly damped Floquet mode near the forcing frequency. The authors propose signal-energy-based detection and a mitigation that re-times workloads to minimize the facility's oscillatory energy.","solution":"Signal-energy-based detection and mitigation that re-times the workloads to minimize the facility's oscillatory energy, validated through PHIL experiments and real-time simulations on a surrogate model of the Saudi Arabian transmission network.","labels":["security","research"],"sourceUrl":"https://arxiv.org/abs/2611.23666v1","publishedAt":"2026-11-18T00:49:44.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"low","attackType":["denial_of_service"],"issueType":"research","affectedPackages":null,"affectedPackageNames":null,"affectedPackageRefs":null,"affectedVendors":[],"affectedVendorsRaw":["large language model (LLM) training workloads"],"classifierModel":"claude-haiku-5-5","classifierPromptVersion":"v4","summaryPromptVersion":"v2","headline":null,"headlinePromptVersion":null,"cvssVector":null,"attackVector":null,"attackComplexity":null,"privilegesRequired":null,"userInteraction":null,"exploitMaturity":null,"epssScore":null,"epssCheckedAt":null,"kevDateAdded":null,"advisoryAliases":null,"affectedPackagesSource":null,"affectedPackagesCheckedAt":null,"patchAvailable":null,"disclosureDate":"2026-11-18T00:49:44.000Z","fixReleasedAt":null,"daysToFix":null,"fixCheckedAt":null,"capecIds":null,"crossRefCount":0,"attackSophistication":"advanced","impactType":["availability"],"aiComponentTargeted":"inference","llmSpecific":true,"classifierConfidence":0.7,"researchCategory":"preprint","atlasIds":null}}