157 matches found
CLEANSTART-2024-EC17082 Use after free in Window Dialogue in Google Chrome prior to 98
Security vulnerability affects the qt5-qtwebengine package. Use after free in Window Dialogue in Google Chrome prior to 98...
HiTMS_steganography
HiTMS: A High-Throughput Multi-Stream Linguistic Steganography Framework Research code for linguistic steganography over LLM chat dialogues , with a single-stream baseline and a batched multi-stream HiTMS protocol that hides several independent secret messages at once and exploits GPU batching fo...
The Model Plants the Trigger: Answer-Side Backdoor Attacks in Multi-Turn Large Language Models
Safety alignment in Large Language Models LLMs remains vulnerable to backdoor attacks. Existing LLM backdoors are almost all input-centric: activation depends on explicit trigger patterns in the user input, so modern guardrails are built to sanitize the input space. We challenge this assumption...
CLEANSTART-2024-UR50012 Use after free in Window Dialogue in Google Chrome prior to 98
Security vulnerability affects the chromium package. Use after free in Window Dialogue in Google Chrome prior to 98...
Does the Unsafe Gradient Survive a Conversation? on the Fragility of Gradient-Based Jailbreak Detection in Multi-Turn Dialogue
Safety-aligned language models are commonly deployed as multi-turn assistants, which lets adversaries spread unsafe intent across several user turns instead of a single prompt. Gradient-based jailbreak detectors such as GradSafe were developed for single prompts: they score an input by the...
Astra Linux – Vulnerability in Chromium
The use of "after free" in Window Dialogs in Google Chrome before version 98.0.4758.80 allowed a remote attacker to potentially exploit heap corruption through a crafted HTML page...
HiTMS: A High-Throughput Multi-Stream Linguistic Steganography Framework
Research code for linguistic steganography over LLM chat dialogues, with a single-stream baseline and a batched multi-stream HiTMS protocol that hides several independent secret messages at once and exploits GPU batching for much higher throughput. A Bob model asks questions; an Alice model...
PYSEC-2026-2558 Langchain-Chatchat Uses a Broken or Risky Cryptographic Algorithm
A flaw has been found in chatchat-space Langchain-Chatchat up to 0.3.1.3. This issue affects the function PIL.Image.tobytes of the file libs/chatchat-server/chatchat/webuipages/dialogue/dialogue.py of the component Vision Chat Paste Image Handler. This manipulation of the argument...
PT-2026-59215
A flaw has been found in chatchat-space Langchain-Chatchat up to 0.3.1.3. This issue affects the function PIL.Image.tobytes of the file libs/chatchat-server/chatchat/webui pages/dialogue/dialogue.py of the component Vision Chat Paste Image Handler. This manipulation of the argument paste...
OpenClaw input validation error vulnerability (CNVD-2026-26890)
OpenClaw is an intelligent artificial assistant open-sourced by the OpenClaw team. OpenClaw has a vulnerability related to input validation. An attacker can exploit this vulnerability to match allowed list entries based on dialogue metadata rather than a stable sender identity. This allows them t...
Hijacking Agent Memory: Stealthy Trojan Attacks through Conversational Interaction
Large language model LLM agents increasingly leverage long term memory to support persistent and autonomous task execution. However, this capability also introduces a new attack surface: memory poisoning, where adversaries can inject malicious information to influence future behavior. Existing...
cltl-dialogue-evaluation (>=0.1.0.dev1, <=0.2.7.dev1), kwja (>=1.0.0, <=1.2.2) +6 more potentially affected by CVE-2026-4372 via transformers (4.19.4)
transformers PYPI version 4.19.4 is affected by a known vulnerability. The following packages have a transitive dependency on transformers and may be impacted: - cltl-dialogue-evaluation =0.1.0.dev1, =1.0.0, =1.5.3, =0.0.20, =1.3.0, =1.10.1 - zensols-spanmatch 0.0.1 Source cves: CVE-2026-4372...
Chatwoot SQL Injection Vulnerability
Chatwoot is an open-source application developed by Chatwoot itself. It serves as an alternative to proprietary solutions such as customer engagement suites, intercom systems, Zendesk, and Salesforce service clouds. Versions of Chatwoot from 2.2.0 to 4.11.2 contained a SQL injection vulnerability...
MAI-2025-0003
The "Foot-In-The-Door" FITD jailbreak represents a sophisticated multi-turn prompt injection attack targeting Large Language Models LLMs. This technique exploits the psychological principle of incremental commitment, gradually escalating malicious requests to circumvent established safety...
EUVD-2026-27390
A flaw has been found in chatchat-space Langchain-Chatchat up to 0.3.1.3. This issue affects the function PIL.Image.tobytes of the file libs/chatchat-server/chatchat/webuipages/dialogue/dialogue.py of the component Vision Chat Paste Image Handler. This manipulation of the argument...
CVE-2026-7845
A flaw has been found in chatchat-space Langchain-Chatchat up to 0.3.1.3. This issue affects the function PIL.Image.tobytes of the file libs/chatchat-server/chatchat/webuipages/dialogue/dialogue.py of the component Vision Chat Paste Image Handler. This manipulation of the argument...
CVE-2026-7845
The CVE-2026-7845 issue affects chatchat-space Langchain-Chatchat (up to 0.3.1.3). It targets the Vision Chat Paste Image Handler: the function PIL.Image.tobytes in libs/chatchat-server/chatchat/webui_pages/dialogue/dialogue.py, via manipulation of paste_image.image_data, leads to the use of a we...
CVE-2026-7845 chatchat-space Langchain-Chatchat Vision Chat Paste Image dialogue.py PIL.Image.tobytes weak hash
A flaw has been found in chatchat-space Langchain-Chatchat up to 0.3.1.3. This issue affects the function PIL.Image.tobytes of the file libs/chatchat-server/chatchat/webuipages/dialogue/dialogue.py of the component Vision Chat Paste Image Handler. This manipulation of the argument...
ContextualJailbreak: Evolutionary Red-Teaming Via Simulated Conversational Priming
Large language models LLMs remain vulnerable to jailbreak attacks that bypass safety alignment and elicit harmful responses. A growing body of work shows that contextual priming, where earlier turns covertly bias later replies, constitutes a powerful attack surface, with hand-crafted multi-turn...
VMware Spring AI Access Control Vulnerability
VMware Spring AI is a development framework by the American company VMware, which integrates artificial intelligence and large language model capabilities into the Spring ecosystem. VMware Spring AI has a access control vulnerability; this vulnerability stems from injecting filter logic through...