Explore primary-source AI security research, evaluated models, attack techniques, and defensive evidence.
985 research findings · 1123 evaluated models
Matches every word across titles, descriptions, sources, affected systems, and models.
API gateways can pool distinct customers into one upstream cache identity. The authors observe cross-customer cache reuse in five deliberately pooled gateway configurations through two provider interfaces.
Source: arXiv
The authors report historical isolation failures in client-held encrypted reasoning blocks across compatible provider API contexts.
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
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
Shell-command screening can miss harmful actions generated by coding agents. CARE combines static checks with optional model review at the command-dispatch boundary.
Source: arXiv
Untrusted issue descriptions and tool responses can redirect privileged coding and tool agents. Twin Agent separates exploration from execution and restricts the information exchanged between them.
Source: arXiv
AgentS4D measures unsafe actions and state changes across complete workspace-agent executions rather than treating task completion or isolated model responses as safety evidence. Its 328 sandboxed cases introduce risky content through user requests, documents, web resources, tools, third-party skills, and persistent memory, then compare the same cases across four agent harnesses and five model backends.
Source: arXiv
OpenSkillRisk evaluates whether agent harnesses safely handle third-party skills that introduce risky behavior through otherwise plausible, benign tasks. The benchmark assembles 263 risky skills from public agent-skill ecosystems and tests three CLI-agent harnesses against seven risk categories using isolated task workspaces, mocked external services, and execution-level evidence.
Source: arXiv
IssueTrojanBench studies indirect prompt injection when a coding agent processes an apparently ordinary software-development issue or related artifact. Starting with six legitimate seed issues from two Python repositories, the authors construct 696 adversarial issue variants spanning four unsafe-action families and six delivery formats, then execute those variants across six agent-model configurations.
Source: arXiv
HANDBOOK.md measures whether an agent can apply detailed organizational rules while completing realistic, multi-step enterprise tasks. The vendor-authored benchmark includes 65 resettable MCP-backed workflows, policy documents of 20 to 124 pages, and 824 deterministic rubric checks covering required decisions, prohibited actions, and final environment state.
Source: arXiv
Entries summarize publicly available primary-source security research. Model names reflect only systems explicitly evaluated by the cited paper, and measurements are research-reported unless independent verification is stated.