J-Anchor: Defending Against Prefill Attacks via J-Space Selective Refusal Restoration

arXiv:2611.01342v1 Announce Type: cross Abstract: Prefill jailbreaks can cause safety-aligned language models to continue an unsafe assistant prefix even when they would refuse the same request under normal decoding. We study how the model's internal harmfulness and refusal signals evolve along this attacker-controlled trajectory. Using the Jacobian lens, we track vocabulary-grounded harmfulness and refusal signals from the assistant header through prefilling and subsequent generation. We identify a characteristic phenomenon, which we call refusal-state erosion: both the refusal signal and the harmfulness signal weaken as the prefill progresses, but the refusal signal progressively declines to the benign baseline by the end of the prefill, while the harmfulness signal remains above its benign baseline and detectable during subsequent generation. Motivated by this observation, we introduce J-Anchor, a training-free inference-time defense that uses the model's pre-attack harmfulness and refusal signals as a request-specific anchor and selectively restores weakened refusal in J-space during prefilling and decoding. J-Anchor substantially reduces attack success under both fixed and deep prefill attacks across different model families. J-Anchor largely preserves benign behavior and reasoning ability, and ablations highlight the importance of targeted restoration in the J-space refusal subspace. Our results suggest that preserving latent refusal dynamics offers a promising route to robust defense against prefill jailbreaks.

Abstract:Prefill jailbreaks can cause safety-aligned language models to continue an unsafe assistant prefix even when they would refuse the same request under normal decoding. We study how the model's internal harmfulness and refusal signals evolve along this attacker-controlled trajectory. Using the Jacobian lens, we track vocabulary-grounded harmfulness and refusal signals from the assistant header through prefilling and subsequent generation. We identify a characteristic phenomenon, which we call refusal-state erosion: both the refusal signal and the harmfulness signal weaken as the prefill progresses, but the refusal signal progressively declines to the benign baseline by the end of the prefill, while the harmfulness signal remains above its benign baseline and detectable during subsequent generation. Motivated by this observation, we introduce J-Anchor, a training-free inference-time defense that uses the model's pre-attack harmfulness and refusal signals as a request-specific anchor and selectively restores weakened refusal in J-space during prefilling and decoding. J-Anchor substantially reduces attack success under both fixed and deep prefill attacks across different model families. J-Anchor largely preserves benign behavior and reasoning ability, and ablations highlight the importance of targeted restoration in the J-space refusal subspace. Our results suggest that preserving latent refusal dynamics offers a promising route to robust defense against prefill jailbreaks.

Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)

arXiv:2611.01342 [cs.CR]

(or arXiv:2611.01342v1 [cs.CR] for this version)

https://doi.org/10.48550/arXiv.2611.01342 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

aiqbench analysis

J-Anchor: Defending Against Prefill Attacks via J-Space Selective Refusal Restoration

Related coverage