AI Threat Alert
Defense MEDIUM
ClawSafety: "Safe" LLMs, Unsafe Agents
Bowen Wei, Yunbei Zhang, Jinhao Pan +5 more
Personal AI agents like OpenClaw run with elevated privileges on users' local machines, where a single successful prompt injection can leak...
AI Security Research
2,583+ academic papers on AI security, attacks, and defenses
Total
2,583
Attack
994
Benchmark
740
Defense
355
Tool
275
Survey
146
Showing 441–460 of 2,583 papers
Attack HIGH
No Attacker Needed: Unintentional Cross-User Contamination in Shared-State LLM Agents
Tiankai Yang, Jiate Li, Yi Nian +5 more
LLM-based agents increasingly operate across repeated sessions, maintaining task states to ensure continuity. In many deployments, a single agent...
Defense MEDIUM
Safety, Security, and Cognitive Risks in World Models
Manoj Parmar
World models -- learned internal simulators of environment dynamics -- are rapidly becoming foundational to autonomous decision-making in robotics,...
Attack HIGH
AgentWatcher: A Rule-based Prompt Injection Monitor
Yanting Wang, Wei Zou, Runpeng Geng +1 more
Large language models (LLMs) and their applications, such as agents, are highly vulnerable to prompt injection attacks. State-of-the-art prompt...
Benchmark MEDIUM
SERSEM: Selective Entropy-Weighted Scoring for Membership Inference in Code Language Models
Kıvanç Kuzey Dikici, Serdar Kara, Semih Çağlar +2 more
As Large Language Models (LLMs) for code increasingly utilize massive, often non-permissively licensed datasets, evaluating data contamination...
Defense MEDIUM
Multi-Agent LLM Governance for Safe Two-Timescale Reinforcement Learning in SDN-IoT Defense
Saeid Jamshidi, Negar Shahabi, Foutse Khomh +2 more
Software-Defined Networking (SDN) is increasingly adopted to secure Internet-of-Things (IoT) networks due to its centralized control and programmable...
Tool HIGH
Automated Framework to Evaluate and Harden LLM System Instructions against Encoding Attacks
Anubhab Sahu, Diptisha Samanta, Reza Soosahabi
System Instructions in Large Language Models (LLMs) are commonly used to enforce safety policies, define agent behavior, and protect sensitive...
Tool HIGH
PDA: Text-Augmented Defense Framework for Robust Vision-Language Models against Adversarial Image Attacks
Jingning Xu, Haochen Luo, Chen Liu
Vision-language models (VLMs) are vulnerable to adversarial image perturbations. Existing works based on adversarial training against task-specific...
Other MEDIUM
SCPatcher: Automated Smart Contract Code Repair via Retrieval-Augmented Generation and Knowledge Graph
Xiaoqi Li, Shipeng Ye, Wenkai Li +1 more
Smart contract vulnerabilities can cause substantial financial losses due to the immutability of code after deployment. While existing tools detect...
Defense LOW
LibScan: Smart Contract Library Misuse Detection with Iterative Feedback and Static Verification
Yishun Wang, Wenkai Li, Xiaoqi Li +3 more
Smart contracts are self-executing programs that manage financial transactions on blockchain networks. Developers commonly rely on third-party code...
Attack HIGH
When Safe Models Merge into Danger: Exploiting Latent Vulnerabilities in LLM Fusion
Jiaqing Li, Zhibo Zhang, Shide Zhou +3 more
Model merging has emerged as a powerful technique for combining specialized capabilities from multiple fine-tuned LLMs without additional training...
Benchmark LOW
BloClaw: An Omniscient, Multi-Modal Agentic Workspace for Next-Generation Scientific Discovery
Yao Qin, Yangyang Yan, Jinhua Pang +1 more
The integration of Large Language Models (LLMs) into life sciences has catalyzed the development of "AI Scientists." However, translating these...
Tool HIGH
The Persistent Vulnerability of Aligned AI Systems
Aengus Lynch
Autonomous AI agents are being deployed with filesystem access, email control, and multi-step planning. This thesis contributes to four open problems...
Benchmark MEDIUM
EnsembleSHAP: Faithful and Certifiably Robust Attribution for Random Subspace Method
Yanting Wang, Jinyuan Jia
Random subspace method has wide security applications such as providing certified defenses against adversarial and backdoor attacks, and building...
Tool HIGH
Architecting Secure AI Agents: Perspectives on System-Level Defenses Against Indirect Prompt Injection Attacks
Chong Xiang, Drew Zagieboylo, Shaona Ghosh +5 more
AI agents, predominantly powered by large language models (LLMs), are vulnerable to indirect prompt injection, in which malicious instructions...
Attack MEDIUM
Performative Scenario Optimization
Quanyan Zhu, Zhengye Han
This paper introduces a performative scenario optimization framework for decision-dependent chance-constrained problems. Unlike classical stochastic...
Survey LOW
PASM: Population Adaptive Symbolic Mixture-of-Experts Model for Cross-location Hurricane Evacuation Decision Prediction
Xiao Qian, Shangjia Dong
Accurate prediction of evacuation behavior is critical for disaster preparedness, yet models trained in one region often fail elsewhere. Using a...
Other MEDIUM
BotVerse: Real-Time Event-Driven Simulation of Social Agents
Edoardo Allegrini, Edoardo Di Paolo, Angelo Spognardi +1 more
BotVerse is a scalable, event-driven framework for high-fidelity social simulation using LLM-based agents. It addresses the ethical risks of studying...
Attack HIGH
Adversarial Prompt Injection Attack on Multimodal Large Language Models
Meiwen Ding, Song Xia, Chenqi Kong +1 more
Although multimodal large language models (MLLMs) are increasingly deployed in real-world applications, their instruction-following behavior leaves...
Attack LOW
AGFT: Alignment-Guided Fine-Tuning for Zero-Shot Adversarial Robustness of Vision-Language Models
Yubo Cui, Xianchao Guan, Zijun Xiong +1 more
Pre-trained vision-language models (VLMs) exhibit strong zero-shot generalization but remain vulnerable to adversarial perturbations. Existing...
Track AI security vulnerabilities in real time
Get breaking CVE alerts, compliance reports (ISO 42001, EU AI Act), and CISO risk assessments for your AI/ML stack.
Start 14-Day Free Trial