Evidence · DiggingBeagle record
Retained recurrent inference state leaks synthetic secrets through state samples and power traces (research paper)
A hybrid language model can leave private attributes in recurrent state after conversation turns, while carefully controlled GPU power measurements reveal a weaker related signal.
- Published
- Nov 11, 2026
- Source role
- primary disclosure
Evidence record
A hybrid language model can leave private attributes in recurrent state after conversation turns, while carefully controlled GPU power measurements reveal a weaker related signal.
Claim-level citations (4)
- supportsRetained recurrent inference state leaks synthetic secrets through state samples and power traces: The recurrent-state study distinguishes host access to a saved model state from a device-level power side channel. Its NVML-related sibling-MIG observation establishes a measurement route under the tested local privileges, but does not show that an unrelated remote inference tenant could obtain the same software state coordinates or recover an actual person's secrets.
Section IV-E, power-acquisition setup, NVML sampling and sibling-MIG accessibility experiment
- supportsRetained recurrent inference state leaks synthetic secrets through state samples and power traces: With 8,192 sampled state coordinates, the experiment infers hidden four-class condition labels at 83.6 to 89.1 percent against a 25 percent chance baseline; the separate power-channel age test has 9.64-year mean absolute error against 11.33 years baseline.
Experiments, state sampling and power-channel performance tables
- supportsRetained recurrent inference state leaks synthetic secrets through state samples and power traces: The researchers compare limited software-state reads, engineered whole-state acquisition and whole-device power observation, using held-out synthetic data to test what remains identifiable after private turns.
Attacker observation models and state-retention method
- supportsRetained recurrent inference state leaks synthetic secrets through state samples and power traces: In a Qwen experiment, 8,192 sampled recurrent coordinates represented 0.174 percent of state but enabled private-condition classification at 83.6 to 89.1 percent versus 25 percent chance on held-out synthetic subjects; power-trace and full-state models require different attacker privileges.
Section IV, Table II Qwen inference from a shared partial observation; separate power and physical-trace studies