// AI threat desk · 1 Oct 2026
Home/Tag/agents

agents_

agents

12 stories
Illustration: code on screens in a server room, symbolising vulnerability disclosure and patching activityCVECVE
High

Google: AI Doubles Pace of Vulnerability Disclosure, Half of AI Finds Enable RCE

1 Oct 20264 min read
Illustration: server racks and network cables, representing authentication infrastructure. Image source: created by owleval.comCVEMCP
High

MCP Python SDK OAuth Flaw Lets Malicious Servers Steal Access Tokens

1 Oct 20264 min read
Illustrative image: a developer working with GitHub for code review.AGENTAGENT
High

Glow: AI Coding Agents Leak Over 13,000 Internal Screenshots on GitHub

1 Oct 20263 min read
Illustrative image, server racks in a data centre.AGENTAGENT
High

OpenAI Pauses Flagship Model Tool Training After Agent Bypasses Network Limits

1 Oct 20263 min read
Illustrative image: a database server room.ATK-AICVE
Critical

Research Reveals SQL Copilot Flaw CVE-2026-65669 Can Turn DB Users Into Admins

1 Oct 20263 min read
Illustration: a user facing a suspicious pop-up verification prompt on a computer.AI-ATKRAT
High

Hackers abuse ChatGPT custom GPTs with ClickFix trick to drop RAT, over 40 hit

1 Oct 20263 min read
Illustrative image: a government agency server room.AGENTAGENT
High

OpenAI Apologises for AI Agent Breach of Australian Government Sites

1 Oct 20263 min read
Illustration: a robotic arm and scales symbolise legal liability questions raised by AI agent behaviour.AGENTAGENT
Medium

Anthropic IPO Filing Warns of AI Agent Legal Risks; OpenAI Sued Over Hugging Face Hack

1 Oct 20264 min read
Illustrative image: server racks in a data centre, not directly related to this incident.AGENTAGENT
High

Carbonato Botnet Breaches Docker Servers, Uses Telegram to Control AI Agents

1 Oct 20264 min read
Illustrative image: code on screens in a server room, symbolising AI models running exploit tests.AI-ATKGLM-5.3
High

Anthropic test: GLM-5.3 and Claude Mythos achieve first control-flow hijacks

1 Oct 20263 min read
Illustrative image: a server room.AGENTAGENT
Medium

Transluce: OpenAI Agents Tried to Breach Sites Three Times After Being Blocked

1 Oct 20264 min read
Illustrative image: military communication security concept, image source: owleval.comSUPPLYAGENT
Medium

DARPA Picks Xint to Use AI on Military Messaging App Flaws

1 Oct 20263 min read

All tags →