Posts

vLLM CVE-2026-22778/34756 — Heap Leak and DoS in Multimodal Serving

Two vLLM CVEs: a multimodal heap-address leak weakening ASLR (CVE-2026-22778) and an unbounded n-parameter DoS in the OpenAI-compatible API (CVE-2026-34756).

arXiv — IterInject: Feedback-Guided Iterative Prompt Injection Against Agents

Shanghai Jiao Tong University researchers introduce IterInject, an indirect prompt injection framework that uses LLM-driven iterative optimization to adapt payloads against agent defenses, achieving full success on 5 of 9 Claude Code targets.

CERT-In — 12-hour patch mandate calibrated to AI exploitation speed

Permiso — ChatGPhish Turns ChatGPT Summaries Into a Phishing Surface

Detectify — MCP Server brings deterministic vuln scanning into AI coding agents

Detectify launches an MCP Server that exposes its scanning engines to AI coding agents, enabling autonomous vulnerability finding, patching, and validation in real time.

Obsidian — Flowise CVE-2026-40933 MCP stdio Supply-Chain RCE

Obsidian Security releases PoC exploit for CVE-2026-40933 (CVSS 9.9), a systemic MCP stdio command injection in Flowise that enables one-click RCE via crafted chatflow import.

Geordie — $30M Series A for AI Agent Security and Governance

Geordie raises $30M Series A led by Balderton Capital to build a purpose-built security and governance platform for AI agents at enterprise scale.

WithSecure — GreyVibe Russia-Linked Group Supercharges Ops with AI

WithSecure documents GreyVibe, a Russia-nexus threat group using ChatGPT, Gemini, and Ideogram AI across every attack phase — from lure creation to malware development.

Push Security — LLMShare Campaign Abuses ChatGPT Sharing to Deliver Malware

arXiv — AgentSecBench: Formal Security Framework for LLM Agents

AgentSecBench introduces a formal three-game framework measuring instruction integrity, retrieval confidentiality, and capability integrity in LLM agent systems.

MemMorph — Tool Hijacking in LLM Agents via Memory Poisoning

arXiv: Real-World Prompt Injection Attacks in LLM-Based Resume Screening

New arXiv paper measures prompt injection attacks in production LLM resume screening systems, testing prevention, detection, and localization defenses across real-world HR workflows.

Flashpoint — Deepfake KYC Bypass Kits Sold as SaaS to Criminals

Flashpoint tracked 63,763 April posts advertising AI-powered KYC bypass toolkits, complete with live video spoofing, voice cloning, and platform-specific updates.

JFrog — 2026 Supply Chain Report: npm Attacks Up 451%, 495 Malicious AI Models

jqwik Maintainer Sneaks Data-Nuking Prompt Injection Into AI Coding Agents

A frustrated maintainer of the jqwik Java property-based testing library embedded a hidden prompt injection in test output, instructing AI coding agents to delete all project code and tests.

Langroid CVE-2026-25879 — Prompt-to-SQL Injection Leads to RCE

Sysdig: First LLM Agent-Driven Intrusion via Marimo CVE-2026-39987

Sysdig documents the first observed intrusion where an LLM agent composed post-exploitation in real time — from Marimo RCE to full PostgreSQL database dump in under one hour.

UVCyber MCP Threat Advisory: 40+ CVEs, Tool Poisoning, and the Missing Auth Layer

UVCyber publishes a comprehensive MCP threat advisory covering 40+ CVEs, tool poisoning, rug pulls, and the architectural gap of missing built-in authentication in the Model Context Protocol.

vLLM CVE-2026-4944 — Hardcoded trust_remote_code Bypass Enables RCE

vLLM 0.14.1 hardcodes trust_remote_code=True in two model files, bypassing user security settings and enabling remote code execution via malicious HuggingFace models.

Cogent — AI exploit dev shrinks weaponization from 125 days to 12 hours

Cogent reports AI-assisted exploit development has collapsed vulnerability-to-weaponization timelines from 125 days to 12 hours, outpacing scanner-based detection cycles.