BASIS uses two linear probes on prefill attention to separately detect injection presence and predict whether the injection would actually compromise the model, refusing only when both fire.
Sandwich defense,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CR 1years
2026 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
BASIS: Breach-Aware Selective Prompt Injection Shielding with Prefill Attention Probes
BASIS uses two linear probes on prefill attention to separately detect injection presence and predict whether the injection would actually compromise the model, refusing only when both fire.