Skip to content

Scan record

Everything the site collected, newest first by collection time (Beijing time), back seven days.

Nov 1738 loaded

  1. Operating points of frozen guardrail probes fail under base model updates

    PaperarXiv: 其他 AI 安全论文The item you came from

  2. On-Policy Distillation as a Backdoor Propagation Channel

    PaperarXiv: 其他 AI 安全论文

  3. A Bag of Concepts? Testing Role-Filler Bindings in the Verbalizable Workspace

    PaperarXiv: 其他 AI 安全论文

  4. Agent-ZTA: A Chain-Aware Zero-Trust Model for Bidirectional Security State Propagation and Runtime Containment in Multi-Agent Delegation

    PaperarXiv: 其他 AI 安全论文

  5. MemeEconomy: Will LLM agents trade ethics for survival?

    PaperarXiv: 其他 AI 安全论文

  6. Same Evidence, Different Verdicts: How Delivery of CT Images and DICOM Metadata Changes VLM Quality Control Decisions

    PaperarXiv: 其他 AI 安全论文

  7. Quantum-Inspired Orthogonal Adapters for Certifiably Robust Medical Image Classification

    PaperarXiv: 其他 AI 安全论文

  8. Not All Steps Matter: Time-Adaptive Spectral Control for Concept Removal in Diffusion Models

    PaperarXiv: 其他 AI 安全论文

  9. Rethinking the Verification of Machine Unlearning

    PaperarXiv: 其他 AI 安全论文

  10. When Helpful Agents Cause Economic Harm: Evaluating the Safety of LLM Agents in Executable Economic Worlds

    PaperarXiv: 其他 AI 安全论文

  11. NYUAD participates in the ArGuard shared task: a multimodal embedding model for detecting Arabic hateful memes and unsafe prompts

    PaperarXiv: 其他 AI 安全论文

  12. Jailbreaking LLMs with language style as a hidden attack surface

    PaperarXiv: 其他 AI 安全论文

  13. Endogenous preferences as vulnerability signals: discovering universal adversarial trigger words that degrade LLM utility

    PaperarXiv: 其他 AI 安全论文

  14. mcp-atlassian fixes a credential misuse vulnerability in which HTTP transport does not validate Host and Origin under specific configurations

    Newssooperset/mcp-atlassianEvent

  15. mcp-atlassian 0.24.5 concurrent Markdown writes may mix in content from other requests, fixed in 0.24.6

    Newssooperset/mcp-atlassianEvent

  16. We Hold All the Cards: Securing Agent Cards at the runtime trust boundary of the Agent2Agent protocol

    PaperarXiv: 其他 AI 安全论文

  17. When the spec sheet lies: injection attacks against LLM-generated compliance documents

    PaperarXiv: 其他 AI 安全论文

  18. Optional Traps: An Empirical Study of LLM Agents Facing Credential Deception and Alert-Driven Revocation

    PaperarXiv: 其他 AI 安全论文

  19. VAST: Compiling Real Vulnerability Scenarios into Verifiable Multi-Step Environments to Evaluate LLM Red-Team Agents

    PaperarXiv: 其他 AI 安全论文

  20. Agent Authenticator: Passkey-level hardware-bound authentication for AI agents

    PaperarXiv: 其他 AI 安全论文

  21. Attacker diversity reduction in adaptive LLM red teaming

    PaperarXiv: 其他 AI 安全论文

  22. Compositional Exploitation in Autonomous Coding Agents: Attack Mechanisms and Ad Hoc Defenses

    PaperarXiv: 其他 AI 安全论文

  23. Security of deep research agents: attack surface, defenses, benchmarks, and agenda seeding targeting research planning

    PaperarXiv: 其他 AI 安全论文

  24. Beyond Patches: Control-Gated Benign Probe Reconstruction of Patch Bypasses from Incomplete Fixes in LLM Agent Frameworks

    PaperarXiv: 其他 AI 安全论文

  25. AudioTrace: File-Level Membership Detection for Audio LLMs

    PaperarXiv: 其他 AI 安全论文

  26. Swarm Shepherd: Protecting Multi-Agent Ecosystems from Persistent Latent Intrusions

    PaperarXiv: 其他 AI 安全论文

  27. German research reveals that at least 7,403 databases of apps developed by users on the AI development platform Lovable have improper permission settings

    NewsiThome (Taiwan)Event

  28. InclusionMed: An AI benchmark for inclusive healthcare

    PaperHugging Face Daily Papers

  29. Belo Horizonte City Council to hold hearing on November 18 to discuss AI in public safety

    NewsPortal CMBHGovernance

  30. DominoBench: evaluating multi-step physical reasoning from static images

    PaperHugging Face Daily Papers

  31. SWE-CoWork: Benchmarking Coding Agents as Digital Employees in Organizations

    PaperHugging Face Daily Papers

  32. Centered Ablation: why last-layer direction ablation is insufficient or collapses, and a closed-form correction

    PaperHugging Face Daily Papers

  33. A quick guide to data center construction ban litigation

    NewsFinancial Times · AIAI updates

  34. OpenAI uses AI to strengthen cyber defense for NATO and allied countries

    NewsOpenAIEvent

  35. According to sources, Nvidia warned chip buyers that remote access to AI chips from China may bring legal risks

    NewsCNBC · TechnologyGovernance

  36. Apollo Research member: training against CoT-like surfaces may be similar to training against CoT

    NewsLessWrongAlignment

  37. Australia and 10 other countries signed the US-led San Francisco Declaration to protect the sensitive technologies and supply chains needed for advanced AI

    NewsACSC (Australia)AI updates

  38. Helen Clark: Automated warfare has arrived, and governments must restrain it

    NewsThe Guardian · AIGovernance