{
  "id": 20324498,
  "title": "Attackers are using AI agents to automate cybercrime at 'industrial scale,' FBI and AWS warn",
  "url": "https://urgent.news/2026/11/16/attackers-are-using-ai-agents-to-automate-cybercrime-at-industrial",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-16T19:52:00.000Z",
  "source": {
    "name": "The Register Science",
    "slug": "the-register-science",
    "url": "https://www.theregister.com/cyber-crime/2026/11/16/attackers-are-using-ai-agents-to-automate-cybercrime-at-industrial-scale-fbi-and-aws-warn/5310223"
  },
  "original_language": "en",
  "account": "Cybercriminals are leveraging open-source AI penetration testing framework ARTEX to automate attacks on critical sectors such as banks, governments, energy, healthcare, and other industries worldwide, as reported by the FBI and Amazon Threat Intelligence. In a span of two weeks, from October 3 to November 5, Amazon’s MadPot honeypot threat sensors identified two ARTEX campaigns. The first campaign involved stealing AI credentials at an unprecedented scale to gain unauthorized access to AI models, while the second campaign generated deepfake videos to bypass identity verification processes.\n\nIn the first campaign, cybercriminals employed AI agents to steal API keys and feed these keys into ARTEX’s model gateway to create additional agents that sought out more AI model access keys. AWS Chief Information Security Officer CJ Moses detailed how these agents validated identities, enumerated storage, assessed permission scope, and targeted managed AI model services, secrets, parameter stores, and application identities, including LLM gateways and proxy platforms like One API, New API, Veloera, Sub2API, and LiteLLM. Additionally, the attackers utilized cloud credentials to gain access to stolen AI models.\n\nThe second campaign saw ARTEX being used to create real-time deepfake videos to trick identity verification checks at two cryptocurrency exchanges. The attackers reverse-engineered one vendor's liveness protocol, documenting the face-detection event stream and tracking facial position throughout the check. Although the agents managed to complete the movement sequence with the face positioned 7 pixels away from the target, the vendor's server-side verdict still rejected the deepfake in every attempt. Furthermore, agents constructed a real-time deepfake video pipeline on a single consumer-grade GPU using an open-source face animation model, achieving a frame rate of 11.5 frames per second, nearing the requirement for live video checks.\n\nWhile the ultimate goal of the campaign was cryptocurrency theft, it highlights the AI agents' ability to work simultaneously on multiple tasks, such as cashing out, clearing account balances, setting up a deepfake video pipeline, and bypassing verification checks. These attack techniques, though not novel, demonstrate how AI agents can expedite and sustain the attacker's efforts. Moses emphasized that the majority of these attack methods stem from fundamental security hygiene issues, such as default or hardcoded credentials, exposed keys in code, public artifacts, unauthenticated interfaces and inference endpoints, and unpatched software. Implementing these basic security measures remains the most effective defense against such attacks.\n\nThe FBI likewise warned of the alarming speed at which ARTEX can chain attack techniques, noting that some intrusions progressed from initial targeting to privileged access in under three hours, as seen in the August South Korean financial institution attacks that compromised personal data of about 67,000 individuals. The FBI also highlighted a September ARTEX campaign targeting ConnectWise Cloud Backup customers in a supply chain attack that ultimately compromised over 2,250 downstream Microsoft tenants. Although the original ARTEX developer removed the GitHub repository in October, forked and localized variations, along with a commercial version, persist. As of November 4, the internet scanning service Censys identified 1,465 hosts running an exposed ARTEX console, with the total number of deployments likely being significantly higher when considering systems behind firewalls or in private networks.",
  "summary": "Critical orgs around the globe under seige",
  "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."
}