AI Threat Alert
Multi-Turn Jailbreaking of Aligned LLMs via Lexical Anchor Tree Search
injection. LATS reformulates jailbreaking as a breadth-first tree search over multi-turn dialogues, where each node incrementally injects missing content words from the attack goal into benign prompts. Evaluations
CoT-Guard: Small Models for Strong Monitoring
attacks, where the adversary is a third-party LLM router injecting hidden objectives into code-generation requests through either prompt manipulation or code manipulation attacks. To push beyond objectives that
Knowledge Base Poisoning Attacks and Defense for Policy-Aware LLM-RAG Framework
operator query types without requiring knowledge of runtime prompts. The attack achieves 85% LLM context corruption from a single injected rule (1.6% poisoning rate) and saturates at 7.7% poisoning, demonstrating
Generate "Normal", Edit Poisoned: Branding Injection via Hint Embedding in Image Editing
rendered onto semantically related objects, even when the user prompt does not explicitly mention it. This form of hidden payload injection makes the attack stealthy. We study two realistic attack
Synthetic Voices, Real Threats: Evaluating Large Text-to-Speech Models in Generating Harmful Audio
second leverages audio-modality exploits (Read, Spell, Phoneme) that inject harmful content through auxiliary audio channels while maintaining benign textual prompts. Through evaluation across five commercial LALMs-based TTS systems
Evaluating Adversarial Vulnerabilities in Modern Large Language Models
prompted to circumvent their own safety protocols, and 'cross-bypass', where one model generated adversarial prompts to exploit vulnerabilities in the other. Four attack methods were employed - direct injection, role
MemVenom: Triggered Poisoning of Multimodal Memories in Web Agents
induction that leverages adversarial perturbations and stealthy OCR injection to override the original user objective. Unlike prior attacks that operate on prompts or text-only memory, our approach enables persistent
Large Language Models for Detecting Cyberattacks on Smart Grid Protective Relays
perfect fault detection accuracy. Additional evaluations demonstrate robustness to prompt formulation variations, resilience under combined time-synchronization and false-data injection attacks, and stable performance under realistic measurement noise levels
Exposing Citation Vulnerabilities in Generative Engines
perspectives of citation publishers and the content-injection barrier, defined as the difficulty for attackers to manipulate answers to user prompts by placing malicious content on the web. GEs integrate
Cross-Cutting Security Analysis of LLM-Generated Code via Metamorphic Testing and Association Rule Mining
level risk factors. We define nine MRs covering major CWE categories, including SQL injection, XSS, command injection, path traversal, hard-coded credentials, weak cryptography, and memory-safety errors, and apply
Zombie Agents: Persistent Control of Self-Evolving LLM Agents via Self-Reinforcing Injections
that memory evolution can convert one-time indirect injection into persistent compromise, which suggests that defenses focused only on per-session prompt filtering are not sufficient for self-evolving agents
Sockpuppetting: Jailbreaking LLMs Without Optimization Through Output Prefix Injection
assistant message block rather than the user prompt, increasing ASR by 64% over GCG on Llama-3.1-8B in a prompt-agnostic setting. The results establish sockpuppetting
Automating Agent Hijacking via Structural Template Injection
ecosystem, enables adversaries to manipulate execution by injecting malicious instructions into retrieved content. Most existing attacks rely on manually crafted, semantics-driven prompt manipulation, which often yields low attack success
Breaking MCP with Function Hijacking Attacks: Novel Threats for Function Calling and Agentic Models
powered system by invoking external functions. Injection and jailbreaking attacks have been extensively explored to showcase the vulnerabilities of LLMs to user prompt manipulation. The expanded capabilities of agentic models
TypePilot: Leveraging the Scala Type System for Secure LLM-generated Code
enforce safety constraints, just as naive prompting for more secure code, our type-focused agentic pipeline substantially mitigates input validation and injection vulnerabilities. The results demonstrate the potential of structured
Reasoning Hijacking: Subverting LLM Classification via Decision-Criteria Injection
which attempts to override the system prompt, Reasoning Hijacking accepts the high-level goal but manipulates the model's decision-making logic by injecting spurious reasoning shortcut. Though extensive experiments
ShadowLogic: Backdoors in Any Whitebox LLM
injecting an uncensoring vector into its computational graph representation. We set a trigger phrase that, when added to the beginning of a prompt into the LLM, applies the uncensoring vector
Differential Harm Propensity in Personalized LLM Agents: The Curious Case of Mental Health Disclosure
benign counterparts) under controlled prompt conditions that vary user-context personalization (no bio, bio-only, bio+mental health disclosure) and include a lightweight jailbreak injection. Our results reveal that harmful
Formal Mechanisms for Market Stability in Self-Interested Agent Societies: A Marketplace Simulation Study
conditions under progressive troll injection over 200 rounds, identifying Mediation as the top-performing mechanism; and (2) adversarial red-teaming of Mediation using iteratively prompt-optimised LLM-driven trolls, finding
P2P: A Poison-to-Poison Remedy for Reliable Backdoor Defense in LLMs
algorithm. P2P injects benign triggers with safe alternative labels into a subset of training samples and fine-tunes the model on this re-poisoned dataset by leveraging prompt-based learning