{
  "id": 20324761,
  "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-20324761",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-16T19:52:00.000Z",
  "source": {
    "name": "The Register",
    "slug": "the-register",
    "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": "Attackers are leveraging open-source AI penetration testing framework ARTEX to orchestrate cybercrime on a massive scale targeting banks, governments, energy, healthcare, and other essential sectors worldwide, according to alerts from the FBI and Amazon Threat Intelligence. From October 3 to November 5, Amazon's MadPot honeypot threat sensors detected two ARTEX campaigns. The first campaign enabled thieves to pilfer AI credentials at \"industrial scale\" granting unauthorized access to models. The second campaign involved generating deepfake videos to deceive live selfie checks. In the first instance, perpetrators utilized AI agents to obtain API keys, feeding these keys into ARTEX's model gateway to generate more agents seeking access to additional AI model keys. These agents validated identities through token and identity services, examined storage, evaluated permission scope, and targeted managed AI model services, secrets, parameter stores, and application identities. Targets included Large Language Model (LLM) gateways and proxy platforms like One API, New API, Veloera, Sub2API, and LiteLLM, as well as AI application builders such as Dify, FastGPT, Langflow, RAGFlow, Open WebUI, LobeChat, and Ollama. The campaign also employed cloud credentials to access stolen models. In the second campaign, ARTEX was used to craft real-time deepfake videos to bypass identity verification (KYC) procedures at two cryptocurrency exchanges and forge ID photos for document checks. Although the agents succeeded in executing the movement sequence within a 7-pixel margin of the target, the server-side verdict still rejected the deepfake in all attempts. Additionally, agents constructed a real-time deepfake video pipeline on a single low-end GPU using an open-source face animation model, achieving a frame rate of 11.5 frames per second, close to the expected live video check. In another session, agents developed synthetic camera metadata to match the real image dimensions and compression tables in JPEG files, thereby circumventing document checks. These attacks aimed to steal cryptocurrency, highlighting ARTEX's capability to execute multiple sessions simultaneously. One agent focused on cashing out, another on clearing account balances, another on setting up the deepfake video pipeline, and another on bypassing identity verification checks. While the techniques used are not novel, these campaigns demonstrate how AI agents can expedite and sustain attacks. AWS Chief Information Security Officer CJ Moses emphasized in a blog post that the absence of basic security hygiene, such as default or hardcoded credentials, keys exposed in code, unauthenticated interfaces and endpoints, and unpatched software, facilitated the success of these AI-driven attacks. The FBI echoed these concerns, stating in a separate alert that ARTEX significantly shortens the time required to chain attack techniques, with some intrusions progressing from initial targeting to privileged access in under three hours. The FBI cited August attacks against South Korean financial institutions that compromised personal data of approximately 67,000 people and a September campaign targeting ConnectWise Cloud Backup customers in a supply chain attack that affected over 2,250 downstream Microsoft tenants. Following the public disclosure of these breaches, the ARTEX developer removed the GitHub repository in October, but forked and localized versions, as well as a commercial version, continue to exist. As of November 4, internet scanning service Censys identified 1,465 hosts running an exposed ARTEX console, though the actual number of deployments is likely significantly higher when including systems behind firewalls or private networks. The FBI urged all organizations, particularly those operating internet-exposed applications, self-service portals, cloud infrastructure, and partner or supplier systems, to implement the mitigation measures outlined in its Flash alert, including reducing external attack surfaces, restricting sensitive services to trusted networks, moving enterprise applications behind zero-trust access policies and permissions, and promptly patching software.",
  "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."
}