Impact
Security concerns affecting AI safety and alignment
798 matching entries
Controlled safety-tuning experiments link boilerplate refusal statements to unnecessary refusals of benign requests. Request-specific rationales improve benign compliance, with benchmark-dependent safety tradeoffs.
Source: arXiv
KoNA measures whether vision-language models answer valid image questions while refusing unsafe components or correcting unsupported premises. Its 9,300 question-answer pairs include mixed and fully answerable controls.
Source: arXiv
A stored interaction can later steer a memory-augmented agent's answer without direct memory-store access. The study evaluates persistent response manipulation in MemoryOS and MemGPT.
Source: arXiv
Lower-trust tool content can assert facts beyond its authority and distort an agent's decisions. PIPES screens response units against source provenance and expected field meaning.
Source: arXiv
SRE-Bench evaluates whether cybersecurity agents can recover the behavior of realistic binary-only software without relying on source-code memorization. The authors construct 19 private programs, 44 anti-analysis primitives, 262 binary instances, and 1,572 deterministic grading tasks covering security-relevant reverse-engineering scenarios.
Source: arXiv
Self-evolving agent-skill systems may promote recurring, attacker-controlled execution records into persistent reusable instructions. The paper evaluates whether poisoned but task-plausible trajectories survive aggregation and appear in generated skill artifacts, using inert canary behaviors and two structurally different skill-evolution pipelines.
Source: arXiv
SkillSentry evaluates third-party agent skills by constructing source-grounded decoy environments and comparing matched executions with and without the tested skill. The method requires completed, observable, skill-attributed side effects rather than treating suspicious text, ordinary privileged operations, or unexecuted paths as proven malicious behavior.
Source: arXiv
Third-party fine-tuning adapters may contain backdoors. Z-PEFT screens adapter weights using spectral features, evaluated on PADBench's 13,300 adapters.
Source: arXiv
VulnGym measures whether coding agents can locate and explain repository-level security vulnerabilities from realistic advisory and source-code context. The benchmark contains 184 reviewed advisories, 408 line-annotated vulnerability entries, and 23 repositories, with separate end-to-end detection and oracle-conditioned localization tasks.
Source: arXiv
MemSecBench follows malicious agent-memory content from initial write through persistence, retrieval, action selection, execution, and attempted selective repair. Its controlled Write–Execute–Forget protocol evaluates 310 human-reviewed cases across two harnesses, four memory backends, three model backends, and seven evidence-gated lifecycle checkpoints.
Source: arXiv