Issues

[new-research] arXiv:2611.06498 — Signed but Steerable: Measuring and Closing the Text-Vector Integrity Gap in Off

GenAI-Security-Project/crosswalk#377 · opened

on Nov 9, 2026

[new-research] arXiv:2611.06553 — Addition by Subtraction: Toward Mitigating Fragile Safety Alignment in Large Lan

GenAI-Security-Project/crosswalk#376 · opened

on Nov 9, 2026

[new-research] arXiv:2611.06585 — Signed, Sealed, Leaked: Trust Beyond Signature for MCP Toolchains

GenAI-Security-Project/crosswalk#375 · opened

on Nov 9, 2026

[new-research] arXiv:2611.07222 — SPILLWAY: Separating Target Admission from Sanitization to Defend Tool-Using Age

GenAI-Security-Project/crosswalk#374 · opened

on Nov 9, 2026

[new-research] arXiv:2611.07483 — Signed but Not Safe: Requirement Integrity in Multi-Agent LLM Handoffs

GenAI-Security-Project/crosswalk#373 · opened

on Nov 9, 2026

[new-research] arXiv:2611.07486 — MEvince: Knowledge-Constrained LLM Reasoning for Technique Identification from M

GenAI-Security-Project/crosswalk#372 · opened

on Nov 9, 2026

[new-research] arXiv:2611.07611 — A Gingerbread House for LLM Agents: Source-Planned Honeypots That Lure Autonomou

GenAI-Security-Project/crosswalk#370 · opened

on Nov 9, 2026

[new-research] arXiv:2611.07530 — JailBound: FOL-Guided Jailbreak Evaluation for Exploring LLM Behavioral Boundari

GenAI-Security-Project/crosswalk#371 · opened

on Nov 9, 2026

[new-research] arXiv:2611.08170 — The Promise of Strategic Agentic Scambaiting

GenAI-Security-Project/crosswalk#368 · opened

on Nov 9, 2026

[new-research] arXiv:2611.07640 — Beyond Instruction Override: Benchmarking Indirect Prompt Injection with Task-Ta

GenAI-Security-Project/crosswalk#369 · opened

on Nov 9, 2026

[new-research] arXiv:2611.08383 — FedReason: Large Language Models as Weight-Space Meta-Aggregators for Heterogene

GenAI-Security-Project/crosswalk#366 · opened

on Nov 9, 2026

[new-research] arXiv:2611.08343 — LABBA: LLM-Assisted Behavioral Biometric Authentication

GenAI-Security-Project/crosswalk#367 · opened

on Nov 9, 2026

[new-research] arXiv:2611.08752 — AttentionSwitch: Jailbreak Defense for LLMs via Joint Learning of Sparse Attenti

GenAI-Security-Project/crosswalk#365 · opened

on Nov 9, 2026

[new-research] arXiv:2611.08811 — I Think, Therefore I Attack: Evaluating Thought Injection Attacks on Reasoning A

GenAI-Security-Project/crosswalk#364 · opened

on Nov 9, 2026

[new-research] arXiv:2611.09113 — Authorization Drift in LLM Agents: Evaluating Boundary Compliance in Multi-Step

GenAI-Security-Project/crosswalk#362 · opened

on Nov 9, 2026

[new-research] arXiv:2611.08845 — Jev as an LLM Safety Guardrail: An Empirical Study of Input Filtering, Selective

GenAI-Security-Project/crosswalk#363 · opened

on Nov 9, 2026

[new-research] arXiv:2611.09607 — ReplayAudit: Auditing Outsourced LLM Fine-Tuning via Sealed Records and Selectiv

GenAI-Security-Project/crosswalk#360 · opened

on Nov 9, 2026

[new-research] arXiv:2611.09446 — Logit Laundering: Evading Data-Use Auditing in Large Language Models

GenAI-Security-Project/crosswalk#361 · opened

on Nov 9, 2026

[new-research] arXiv:2611.09721 — MemoryDojo: Benchmarking Poisoning Through LLM Agent Memory

GenAI-Security-Project/crosswalk#359 · opened

on Nov 9, 2026

[new-research] arXiv:2611.09734 — GuardrailAgent: An Agentic Framework for Jailbreak Defense

GenAI-Security-Project/crosswalk#358 · opened

on Nov 9, 2026

[framework-update] MITRE ATLAS — content change detected

GenAI-Security-Project/crosswalk#357 · opened

on Nov 2, 2026

[new-cve] CVE-2026-117574 — A flaw was found in ai-gateway-payload-processing. An authenticated remote attac

GenAI-Security-Project/crosswalk#356 · opened

on Nov 2, 2026

[new-cve] CVE-2026-116691 — MLflow is an open source AI engineering platform for agents, large language mode

GenAI-Security-Project/crosswalk#354 · opened

on Nov 2, 2026

[new-cve] CVE-2026-116692 — MLflow is an open source AI engineering platform for agents, large language mode

GenAI-Security-Project/crosswalk#355 · opened

on Nov 2, 2026

[new-cve] CVE-2026-116690 — MLflow is an open source AI engineering platform for agents, large language mode

GenAI-Security-Project/crosswalk#353 · opened

on Nov 2, 2026

Navigation Menu

Repository navigation

Issues

[new-research] arXiv:2611.06498 — Signed but Steerable: Measuring and Closing the Text-Vector Integrity Gap in Off

[new-research] arXiv:2611.06553 — Addition by Subtraction: Toward Mitigating Fragile Safety Alignment in Large Lan

[new-research] arXiv:2611.06585 — Signed, Sealed, Leaked: Trust Beyond Signature for MCP Toolchains

[new-research] arXiv:2611.07222 — SPILLWAY: Separating Target Admission from Sanitization to Defend Tool-Using Age

[new-research] arXiv:2611.07483 — Signed but Not Safe: Requirement Integrity in Multi-Agent LLM Handoffs

[new-research] arXiv:2611.07486 — MEvince: Knowledge-Constrained LLM Reasoning for Technique Identification from M

[new-research] arXiv:2611.07611 — A Gingerbread House for LLM Agents: Source-Planned Honeypots That Lure Autonomou

[new-research] arXiv:2611.07530 — JailBound: FOL-Guided Jailbreak Evaluation for Exploring LLM Behavioral Boundari

[new-research] arXiv:2611.08170 — The Promise of Strategic Agentic Scambaiting

[new-research] arXiv:2611.07640 — Beyond Instruction Override: Benchmarking Indirect Prompt Injection with Task-Ta

[new-research] arXiv:2611.08383 — FedReason: Large Language Models as Weight-Space Meta-Aggregators for Heterogene

[new-research] arXiv:2611.08343 — LABBA: LLM-Assisted Behavioral Biometric Authentication

[new-research] arXiv:2611.08752 — AttentionSwitch: Jailbreak Defense for LLMs via Joint Learning of Sparse Attenti

[new-research] arXiv:2611.08811 — I Think, Therefore I Attack: Evaluating Thought Injection Attacks on Reasoning A

[new-research] arXiv:2611.09113 — Authorization Drift in LLM Agents: Evaluating Boundary Compliance in Multi-Step

[new-research] arXiv:2611.08845 — Jev as an LLM Safety Guardrail: An Empirical Study of Input Filtering, Selective

[new-research] arXiv:2611.09607 — ReplayAudit: Auditing Outsourced LLM Fine-Tuning via Sealed Records and Selectiv

[new-research] arXiv:2611.09446 — Logit Laundering: Evading Data-Use Auditing in Large Language Models

[new-research] arXiv:2611.09721 — MemoryDojo: Benchmarking Poisoning Through LLM Agent Memory

[new-research] arXiv:2611.09734 — GuardrailAgent: An Agentic Framework for Jailbreak Defense

[framework-update] MITRE ATLAS — content change detected

[new-cve] CVE-2026-117574 — A flaw was found in ai-gateway-payload-processing. An authenticated remote attac

[new-cve] CVE-2026-116691 — MLflow is an open source AI engineering platform for agents, large language mode

[new-cve] CVE-2026-116692 — MLflow is an open source AI engineering platform for agents, large language mode

[new-cve] CVE-2026-116690 — MLflow is an open source AI engineering platform for agents, large language mode