Issues

[new-cve] CVE-2026-111079 — A flaw was found in Goose. When the terminal integration feature is enabled, the

GenAI-Security-Project/crosswalk#319 · opened

on Oct 26, 2026

[new-cve] CVE-2026-115886 — MLflow is an open source AI engineering platform for agents, large language mode

GenAI-Security-Project/crosswalk#320 · opened

on Oct 26, 2026

[new-research] arXiv:2610.30601 — HiMark: Hierarchical and Controllable Multi-bit Watermarking for Large Language

GenAI-Security-Project/crosswalk#318 · opened

on Oct 26, 2026

[new-research] arXiv:2610.30666 — Your Token Is Gone: A Token-Consumption Countermeasure for Disrupting LLM-Based

GenAI-Security-Project/crosswalk#316 · opened

on Oct 26, 2026

[new-research] arXiv:2610.30626 — Model-intrinsic, Weight-level, and Training-free Spectral Fingerprinting for Lar

GenAI-Security-Project/crosswalk#317 · opened

on Oct 26, 2026

[new-research] arXiv:2610.30738 — AgentTriage: Replayable Security Diagnosis Across Heterogeneous Agent Runtimes

GenAI-Security-Project/crosswalk#314 · opened

on Oct 26, 2026

[new-research] arXiv:2610.31713 — Omni-UBench: Benchmarking Omni-Modal LLM-based Guardrails Across Safety, Utility

GenAI-Security-Project/crosswalk#315 · opened

on Oct 26, 2026

[new-research] arXiv:2610.30963 — ReDence: A Hybrid Agentic and Decision-Model Framework for Rapid, Evidence-Groun

GenAI-Security-Project/crosswalk#312 · opened

on Oct 26, 2026

[new-research] arXiv:2610.30798 — Amplifying Latency and Cost of LLM Inference via Rowhammer Attack

GenAI-Security-Project/crosswalk#313 · opened

on Oct 26, 2026

[new-research] arXiv:2610.31404 — AgentPhishing: A Benchmark of LLM Email Agents Under Adaptive Attack

GenAI-Security-Project/crosswalk#310 · opened

on Oct 26, 2026

[new-research] arXiv:2610.31366 — AfterDeleteBench: Auditing Post-Deletion Information Persistence in LLM Agents

GenAI-Security-Project/crosswalk#311 · opened

on Oct 26, 2026

[new-research] arXiv:2610.31914 — Incomparable Minima in Agent Tool Authorization

GenAI-Security-Project/crosswalk#308 · opened

on Oct 26, 2026

[new-research] arXiv:2610.31573 — Social-HUMINT: An LLM-Augmented Unified Framework for Measuring Social Media Exp

GenAI-Security-Project/crosswalk#309 · opened

on Oct 26, 2026

[new-research] arXiv:2610.31995 — TRACE: Data-Free Text Reconstruction Attacks against Approximate Unlearning in L

GenAI-Security-Project/crosswalk#306 · opened

on Oct 26, 2026

[new-research] arXiv:2610.31974 — Agentic AI Defense: Decode-Time Provenance Binding for MCP

GenAI-Security-Project/crosswalk#307 · opened

on Oct 26, 2026

[new-research] arXiv:2610.32079 — Label-Budgeted Attribute Inference from Protected Features in Edge-Cloud Split M

GenAI-Security-Project/crosswalk#304 · opened

on Oct 26, 2026

[new-research] arXiv:2610.32017 — SynMark: Semantic-Coordinate Multi-Bit Watermarking during LLM Code Generation

GenAI-Security-Project/crosswalk#305 · opened

on Oct 26, 2026

[new-research] arXiv:2610.32155 — Detecting Insecure Compositions of Clean Agent Skills via Capability-Flow Graph

GenAI-Security-Project/crosswalk#302 · opened

on Oct 26, 2026

[new-research] arXiv:2610.32092 — AgentTrace: Execution-Level Fingerprinting of Black-Box LLM Agents

GenAI-Security-Project/crosswalk#303 · opened

on Oct 26, 2026

[new-research] arXiv:2610.32531 — On Task, Off Track: Benchmarking LLM and Agent Robustness to Near-Target Prompt

GenAI-Security-Project/crosswalk#300 · opened

on Oct 26, 2026

[new-research] arXiv:2610.32345 — DeepAVR: Deep Analysis via Multi-Agent Collaboration for Automated Vulnerability

GenAI-Security-Project/crosswalk#301 · opened

on Oct 26, 2026

[new-research] arXiv:2610.32876 — Trusting the Tool: Asymmetric Deference of LLM Vulnerability Auditors to Static-

GenAI-Security-Project/crosswalk#299 · opened

on Oct 26, 2026

[Framework Submission] ISO 27090

GenAI-Security-Project/crosswalk#298 · opened

on Oct 26, 2026

[Framework Submission] the lucas robert custom grahpic desgin

GenAI-Security-Project/crosswalk#297 · opened

on Oct 22, 2026

[new-research] arXiv:2610.22578 — Model-as-a-Trojan: Self-Activating Covert DNS Exfiltration Baked Into the Weight

GenAI-Security-Project/crosswalk#295 · opened

on Oct 19, 2026

Navigation Menu

Repository navigation

Issues

[new-cve] CVE-2026-111079 — A flaw was found in Goose. When the terminal integration feature is enabled, the

[new-cve] CVE-2026-115886 — MLflow is an open source AI engineering platform for agents, large language mode

[new-research] arXiv:2610.30601 — HiMark: Hierarchical and Controllable Multi-bit Watermarking for Large Language

[new-research] arXiv:2610.30666 — Your Token Is Gone: A Token-Consumption Countermeasure for Disrupting LLM-Based

[new-research] arXiv:2610.30626 — Model-intrinsic, Weight-level, and Training-free Spectral Fingerprinting for Lar

[new-research] arXiv:2610.30738 — AgentTriage: Replayable Security Diagnosis Across Heterogeneous Agent Runtimes

[new-research] arXiv:2610.31713 — Omni-UBench: Benchmarking Omni-Modal LLM-based Guardrails Across Safety, Utility

[new-research] arXiv:2610.30963 — ReDence: A Hybrid Agentic and Decision-Model Framework for Rapid, Evidence-Groun

[new-research] arXiv:2610.30798 — Amplifying Latency and Cost of LLM Inference via Rowhammer Attack

[new-research] arXiv:2610.31404 — AgentPhishing: A Benchmark of LLM Email Agents Under Adaptive Attack

[new-research] arXiv:2610.31366 — AfterDeleteBench: Auditing Post-Deletion Information Persistence in LLM Agents

[new-research] arXiv:2610.31914 — Incomparable Minima in Agent Tool Authorization

[new-research] arXiv:2610.31573 — Social-HUMINT: An LLM-Augmented Unified Framework for Measuring Social Media Exp

[new-research] arXiv:2610.31995 — TRACE: Data-Free Text Reconstruction Attacks against Approximate Unlearning in L

[new-research] arXiv:2610.31974 — Agentic AI Defense: Decode-Time Provenance Binding for MCP

[new-research] arXiv:2610.32079 — Label-Budgeted Attribute Inference from Protected Features in Edge-Cloud Split M

[new-research] arXiv:2610.32017 — SynMark: Semantic-Coordinate Multi-Bit Watermarking during LLM Code Generation

[new-research] arXiv:2610.32155 — Detecting Insecure Compositions of Clean Agent Skills via Capability-Flow Graph

[new-research] arXiv:2610.32092 — AgentTrace: Execution-Level Fingerprinting of Black-Box LLM Agents

[new-research] arXiv:2610.32531 — On Task, Off Track: Benchmarking LLM and Agent Robustness to Near-Target Prompt

[new-research] arXiv:2610.32345 — DeepAVR: Deep Analysis via Multi-Agent Collaboration for Automated Vulnerability

[new-research] arXiv:2610.32876 — Trusting the Tool: Asymmetric Deference of LLM Vulnerability Auditors to Static-

[Framework Submission] ISO 27090

[Framework Submission] the lucas robert custom grahpic desgin

[new-research] arXiv:2610.22578 — Model-as-a-Trojan: Self-Activating Covert DNS Exfiltration Baked Into the Weight