Showing cs.CRShow all
2 papers · 1 filter
cs.CR2026
From Detection to Decision: Calibration and Deferral for Prompt-Injection Detection
Qianlong Lan, Anuj Kaul, Vinothini Pandurangan +1
Prompt-injection defenses are commonly evaluated as binary detectors, although deployed LLM applications must make operational decisions about whether to allow, block, or escalate…
cs.CR2026
Beyond F1: Evaluating Coverage and Failure Recovery in AI Model Security Scanners
Qianlong Lan, Vinothini Pandurangan, Anuj Kaul +1
Static scanners are increasingly used to identify executable or otherwise unsafe content in machine- learning artifacts, yet conventional evaluation metrics characterize only cases…