Almanac
paper

Can LLMs Reliably Self-Report Adversarial Prefills, and How?

paperactiveprovisionalcan-llms-reliably-self-report-adversarial-prefills-and-how--afaa79bd·1 events·first seen 36h ago

Aliases: Can LLMs Reliably Self-Report Adversarial Prefills, and How?

Co-occurring entities

More like this (12)

Recent events (1)

6arXiv · cs.CL·36h ago·source ↗

LLMs fail to reliably self-report adversarial prefill attacks, study finds

A new arXiv paper evaluates whether LLMs can recognize that their own prior responses were elicited by adversarial prefill attacks, testing ten open-weight models (3B–70B) across four safety benchmarks. Models claim intent on prefilled responses only 27.3% of the time on average, and introspective signal is largely mediated by refusal-related reasoning. Three LoRA fine-tuning methods (SFT, GRPO, DPO) improve the intention-probe gap but counterintuitively raise attack success rates on most models, suggesting partial and fragile mitigation. The findings raise concerns about the reliability of LLM self-reports in safety-critical contexts.