PT-2026-39690

OpenClaw before 2026.4.20 contains a guard bypass vulnerability in the agent-facing gateway config.patch and config.apply endpoints that fails to protect operator-trusted settings including sandbox policy, plugin enablement, gateway auth/TLS, hook routing, MCP server configuration, SSRF policy, a...

CVE-2026-45001: Missing Authorization

OpenClaw before 2026.4.20 contains a guard bypass vulnerability in the agent-facing gateway config.patch and config.apply endpoints that fails to protect operator-trusted settings including sandbox policy, plugin enablement, gateway auth/TLS, hook routing, MCP server configuration, SSRF policy, a...

Comment and Control: Hijacking Agentic Workflows Via Context-Grounded Evolution

Automation platforms such as GitHub Actions and n8n are increasingly adopting so-called agentic workflows, which integrate Large Language Model LLM agents for tasks such as code review and data synchronization. While bringing convenience for developers, this integration exposes a new risk: An...

PT-2026-39688

OpenClaw before 2026.4.20 fails to properly preserve untrusted labels for isolated cron awareness events, allowing webhook-triggered cron agent output to be recorded as trusted system events. Attackers can exploit this trust-labeling issue to strengthen prompt-injection attacks by rendering...

CVE-2026-44999: Insufficient Verification of Data Authenticity

OpenClaw before 2026.4.20 fails to properly preserve untrusted labels for isolated cron awareness events, allowing webhook-triggered cron agent output to be recorded as trusted system events. Attackers can exploit this trust-labeling issue to strengthen prompt-injection attacks by rendering...

OpenClaw 数据伪造问题漏洞

OpenClaw is an open-source intelligent artificial assistant developed by OpenClaw. Versions of OpenClaw prior to 2026.4.20 had a data falsification vulnerability. This vulnerability stemmed from the failure to properly retain the non-trustworthy tags associated with isolated cron events, allowing...

Guaranteed Jailbreaking Defense Via Disrupt-And-Rectify Smoothing

This paper proposes a guaranteed defense method for large language models LLMs to safeguard against jailbreaking attacks. Drawing inspiration from the denoised-smoothing approach in the adversarial defense domain, we propose a novel smoothing-based defense method, termed Disrupt-and-Rectify...

OpenClaw 安全漏洞

OpenClaw is an open-source intelligent artificial assistant developed by OpenClaw. Versions of OpenClaw prior to 2026.4.20 contained security vulnerabilities. These vulnerabilities stemmed from security bypasses in the proxy’s config.patch and config.apply endpoints, which failed to protect...

When Prompts Become Payloads: A Framework for Mitigating SQL Injection Attacks in Large Language Model-Driven Applications

Natural language interfaces to structured databases are becoming increasingly common, largely due to advances in large language models LLMs that enable users to query data using conversational input rather than formal query languages such as SQL. While this paradigm significantly improves usabili...

PT-2026-41181

Name of the Vulnerable Software and Affected Versions Open WebUI versions prior to 0.8.9 Description When a non-administrative user logs into the application, a web request to the '/api/models?' endpoint is initiated. The response from this request reveals the system prompts of available models...

AgentShield: Deception-Based Compromise Detection for Tool-Using LLM Agents

Defenses against indirect prompt injection IPI in tool-using LLM agents share two structural weaknesses. First, they all attempt to prevent attacks rather than detect the compromises that slip through. Second, they have only been evaluated in English, leaving users of low-resource languages such ...

PT-2026-41189

Name of the Vulnerable Software and Affected Versions Open WebUI versions prior to 0.9.5 Description An issue exists in the self-hosted artificial intelligence platform where users can access the system prompt of a model they do not own if a group has been granted read access. This allows for...

Remote Code Execution (RCE)

LiteLLM is vulnerable to Remote Code Execution RCE. The vulnerability is due to unsafe rendering of user-supplied prompt templates in the POST /prompts/test endpoint without sandboxing, allowing authenticated users to execute arbitrary code within the LiteLLM Proxy process and potentially access...

MT-JailBench: A Modular Benchmark for Understanding Multi-Turn Jailbreak Attacks

Multi-turn jailbreaks exploit the ability of large language models to accumulate and act on conversational context. Instead of stating a harmful request directly, an attacker can gradually steer the conversation toward an unsafe answer. Recent methods demonstrate this risk, but they are usually...

Deserialization of Untrusted Data

Overview langchain-core is a Building applications with LLMs through composability Affected versions of this package are vulnerable to Deserialization of Untrusted Data in the load process. An attacker can instantiate trusted classes with untrusted constructor arguments by submitting specially...

banks has Critical Remote Code Execution (RCE) via Jinja2 SSTI

Summary banks = 2.4.1 uses jinja2.Environment unsandboxed to render prompt templates. Applications that pass user-supplied strings as the template argument to Prompt are vulnerable to Server-Side Template Injection SSTI, which can lead to Remote Code Execution RCE on the host system. This is a...

GHSA-GPHH-9Q3H-JGPP banks has Critical Remote Code Execution (RCE) via Jinja2 SSTI

Summary banks = 2.4.1 uses jinja2.Environment unsandboxed to render prompt templates. Applications that pass user-supplied strings as the template argument to Prompt are vulnerable to Server-Side Template Injection SSTI, which can lead to Remote Code Execution RCE on the host system. This is a...

TCLBANKER Banking Trojan Targets Financial Platforms via WhatsApp and Outlook Worms

Threat hunters have flagged a previously undocumented Brazilian banking trojan dubbed TCLBANKER that's capable of targeting 59 banking, fintech, and cryptocurrency platforms. The activity is being tracked by Elastic Security Labs under the moniker REF3076. The malware family is assessed to be a...

CVE-2026-42203

LiteLLM is a proxy server AI Gateway to call LLM APIs in OpenAI or native format. From version 1.80.5 to before version 1.83.7, the POST /prompts/test endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the...

CVE-2026-42261

PromptHub is an all-in-one AI toolbox for prompt, skill, and agent management. From version 0.4.9 to before version 0.5.4, apps/web/src/routes/skills.ts exposes an authenticated endpoint POST /api/skills/fetch-remote that fetches a user-supplied URL server-side and reflects the response body up t...