{
  "id": 20287609,
  "title": "Google: AI agent turned a security patch into a working exploit with no human help",
  "url": "https://urgent.news/2026/11/16/google-ai-agent-turned-a-security-patch-into-a-working-exploit-with-20287609",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-16T14:00:00.000Z",
  "source": {
    "name": "The Register",
    "slug": "the-register",
    "url": "https://www.theregister.com/research/2026/11/16/google-ai-agent-turned-a-security-patch-into-a-working-exploit-with-no-human-help/5309926"
  },
  "original_language": "en",
  "account": "A suspected Russian-speaking attacker employed an AI agent to convert security patches into functional exploits without human intervention, marking a groundbreaking development in autonomous vulnerability discovery and exploitation, according to the Google Threat Intelligence Group (GTIG) in their fourth-quarter 2026 AI threat tracker. This attack, uncovered by threat hunters, involved an exposed staging server hosting a multi-agent vulnerability research and weaponization framework named \"exploit_agent.\" The framework utilized Russian-language tooling and focused on continuous differential analysis of popular web applications, comparing code before and after security updates to identify vulnerabilities, develop exploits, and execute them automatically. This server also housed data stolen from targeted organizations, along with multiple post-compromise scripts enabling activities such as internal reconnaissance, brute-forcing, installing webshells, and removing other attackers' webshells. The operator was particularly interested in targeting French websites in the media, entertainment, and education sectors. The report highlights that attackers are moving away from manual LLM prompting in cyberattacks, deploying fully autonomous, multi-agent offensive security frameworks for multi-stage intrusions. These frameworks have been adopted by both financially motivated cybercriminals and cyber-espionage groups, including those with suspected connections to the People's Republic of China (PRC). Examples of such AI-powered offensive frameworks include ARTEX, CAIRN, CyberStrikeAI, HexStrikeAI, and PentAGI. In another instance, GTIG observed attackers leveraging PentAGI in conjunction with proxy services to target the aerospace sector. The Google researchers emphasize the low barriers to entry, ease of integrating new exploits, and massive scaling potential of these frameworks, leading them to believe that threat actors will likely continue using them in global campaigns targeting organizations.",
  "summary": "Look ma, no human in the loop!",
  "key_points": [],
  "editors_take": null,
  "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."
}