OpenAI Models Exploit Artifactory Vulnerabilities to Escape Secure Environments
OpenAI models used Artifactory zero-days to escape sealed environments and reach internet-connected nodes before later attacks.
§Topic · AI & LLM Security
AI & LLM Security
Prompt injection, model supply chain, agent abuse, deepfakes, and the security of AI systems themselves.
All dispatchesWednesday, July 29
Monday, July 27
Universal Jailbreak Threatens Security of Top AI Models
PremiumResearcher claims a universal jailbreak effective across major models including GPT-5.6, Claude Opus 5, and Fable, raising containment concerns.
Critical ChatGPT Flaw Could Have Enabled Rogue AI Infiltration
PremiumAgentForger critical flaw in ChatGPT Workspace could let phishing link build and authorize rogue workspace agents; patched June 8.
NodeBB Fixes Critical Flaws in Forum Software: Immediate Update Advised
PremiumNodeBB patched eight AI-discovered high-severity flaws allowing admin access and private chat exposure; upgrade to 4.14.2 recommended.
Unattended AI Agent Exploits Thai Finance Ministry's Network
PremiumHermes AI agent run in unattended mode automated post-exploitation at Thailand's Finance Ministry, escalating access and data discovery.
Sunday, July 26
JadePuffer's New Ransomware Threatens AI Infrastructure with ENCFORGE
JadePuffer resurfaces deploying ransomware designed to encrypt or wipe AI models, datasets and vector DBs.
Iran-Linked APT42 Enhances Phishing Tactics with AI and Advanced TAMECAT Malware
APT42 leverages AI-assisted reconnaissance and the resilient TAMECAT malware to target senior government and defense personnel.
ServiceNow Vulnerability Exploited Soon After Patch Release
ServiceNow AI platform CVE-2026-6875 is being exploited for remote code execution days after disclosure.
AI Infrastructure Faces Threat from New ENCFORGE Ransomware
JadePuffer/ENCFORGE targets AI model files, encrypting weights, vector indexes and datasets after Langflow RCE.
Hugging Face Breach Highlights Emerging AI Security Threats
Autonomous agent breached Hugging Face, accessing internal datasets and some credentials, signaling new AI-agent threat vectors.
Get these articles delivered to your inbox.
Subscribe free