SOURCE-LINKED INTELLIGENCE

ServeGuard: Verifiable, Bounded-Residual Confinement of Operator-Invisible Channels Without Revealing the Certified Read Factor

arXiv · AI, language, vision and robotics · article · Sep 18, 2026 · UTC

Third-party adapters for open-weight language models ship as opaque weight matrices; a recipient cannot check whether an adapter hides a backdoor without trusting the publisher or inspecting the weights, the publisher's core asset. For one important class (payloads placed where a safety monitor is structurally blind), detection is unsound as a defense: every detector that factors through the declared monitor is invariant on its blind subspace, and honest and backdoored adapters overlap on every blind-subspace statistic we evaluate, because benign adaptation uses that subspace too. Rather than

Read original source ↗ Open in workspace

recordType

paper

region

Global