{
  "id": 20547249,
  "title": "Google catches AI agents hacking with no human in the loop",
  "url": "https://urgent.news/2026/11/18/google-catches-ai-agents-hacking-with-no-human-in-the-loop",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-18T04:36:31.000Z",
  "source": {
    "name": "e27",
    "slug": "e27",
    "url": "https://e27.co/google-catches-ai-agents-hacking-with-no-human-in-the-loop-20261118/"
  },
  "original_language": "en",
  "account": "Google reveals AI agents are autonomously conducting cyberattacks with no human assistance. The company's Threat Intelligence Group reports the use of AI in attacks has reached a critical inflection point, with attackers deploying multi-agent frameworks for full automation of intrusions. Southeast Asian organizations are targeted in various campaigns, including AI compute theft, phishing kits, and attacks against crypto firms. Open-source \"agentic pentesting\" frameworks like ARTEX, CAIRN, CyberStrikeAI and PentAGI are being exploited by threat actors for malicious purposes. These tools allow attackers to find vulnerabilities, generate payloads and adjust tactics in real-time without human intervention. A custom version of PentAGI was used against US aerospace targets in Asia, while a group with China links built its own automated attack framework for espionage. The report highlights how these AI-driven attacks move quickly, generating more vulnerabilities and exploits between January and August 2026, with zero-day exploits increasing rapidly. Additionally, a new threat called LLMJacking emerges, where attackers exploit AI cloud credits by breaking into companies' cloud accounts and using AI model quotas for their own purposes.",
  "summary": "The hacker hunched over a keyboard is becoming a supervisor. According to Google, more and more of the actual break-in work is now being done by software agents that scan, exploit and adjust on their own. In its latest AI Threat Tracker, published early this week, Google Threat Intelligence Group (GTIG) and its incident response […] The post Google catches AI agents hacking with no human in the…",
  "key_points": [],
  "editors_take": "The shift toward autonomous multi-agent frameworks allows threat actors to execute rapid, real-time intrusions that bypass traditional human-speed defenses and accelerate the deployment of zero-day exploits.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}