Today's cybersecurity headlines are brought to you by ThreatPerspective


Ethical Hacking News

Amazon Uncovers Sophisticated AI Agent Campaign Targeting AI Apps for API Keys




Amazon's threat intelligence team has uncovered a sophisticated AI agent campaign targeting various AI apps for API keys. The campaign, which involved the use of an AI agent called ARTEX, was designed to steal AI provider keys stored inside internet-exposed AI apps. The agents were able to exploit known entry points, use AI-driven social engineering, and target multiple AI apps and services. This campaign highlights the importance of defenders to take steps to secure their AI apps and services, and assumes that exposed secrets will be found quickly.

  • A sophisticated AI agent campaign, ARTEX, targeted various AI apps for API keys, designed to steal AI provider keys from internet-exposed apps.
  • The agents exploited known entry points, used AI-driven social engineering, and targeted multiple AI apps and services.
  • The campaign highlights the importance of treating AI apps, model gateways, and inference endpoints like other production systems.
  • Defenders should change default administrator credentials, search for exposed secrets, and rotate exposed credentials quickly.
  • The campaign is a wake-up call for defenders to secure their AI apps and services, and handle exposed secrets quickly and securely.



  • Amazon's threat intelligence team has uncovered a sophisticated AI agent campaign targeting various AI apps for API keys. The campaign, which involved the use of an AI agent called ARTEX, was designed to steal AI provider keys stored inside internet-exposed AI apps. The agents, which were found to be hunting for weak spots in AI apps such as LiteLLM, One API, and Dify, were able to extract the keys and use them to run AI models that powered the search for more keys.

    According to the records captured by Amazon's threat intelligence team, the agents were able to use the AI models to identify vulnerabilities in the targeted AI apps. The agents then used this information to launch attacks on the apps, exploiting known entry points such as default administrator credentials, unfinished installs, and known flaws in Langflow and RAGFlow.

    The agents were also able to use internet scanning services and certificate transparency logs to search for exposed keys. They were able to search GitHub repositories and event archives, npm and PyPI packages, container image layers, notebooks, model-hosting spaces, paste sites, and Telegram channels for exposed keys.

    The campaign was designed to target AI apps that hold keys for many providers, because one weak admin page there can expose many accounts. The agents focused on apps that have keys for many providers, because one weak admin page there can expose many accounts.

    The playbook of the agents included a list of known entry points for each targeted product, including default administrator credentials such as LiteLLM's default master key, unfinished installs that an outsider could take over, and known flaws in Langflow and RAGFlow. The agents also used server-side request forgery (SSRF) through URL-fetch features to reach cloud metadata services.

    The agents were also able to use AI-driven social engineering to gain access to cloud accounts. The agents were able to create deepfakes of ID documents to bypass identity checks at two cryptocurrency exchanges.

    The campaign was not limited to just one AI app, but rather targeted multiple AI apps and services, including ARTEX, One API, Veloera, Sub2API, LiteLLM, Dify, FastGPT, Langflow, RAGFlow, Open WebUI, LobeChat, and Ollama.

    Amazon's threat intelligence team has notified AWS customers who may have been affected by the campaign. The team has also recommended that defenders check for the presence of ARTEX in their systems and watch for misuse through unexpected spending or unusual usage.

    The team has also recommended that defenders treat AI apps, model gateways, and inference endpoints like other production systems, assuming that exposed secrets will be found quickly. They have also recommended that defenders change default administrator credentials, search configuration files, environment variables, deployment scripts, CI/CD pipelines, and databases for embedded secrets, and rotate exposed credentials quickly.

    The campaign is a wake-up call for defenders to take steps to secure their AI apps and services. It highlights the importance of treating AI apps, model gateways, and inference endpoints like other production systems, and assuming that exposed secrets will be found quickly.

    In conclusion, Amazon's threat intelligence team has uncovered a sophisticated AI agent campaign targeting various AI apps for API keys. The campaign, which involved the use of an AI agent called ARTEX, was designed to steal AI provider keys stored inside internet-exposed AI apps. The agents were able to exploit known entry points, use AI-driven social engineering, and target multiple AI apps and services.

    The campaign highlights the importance of defender's to take steps to secure their AI apps and services. It is a reminder that AI apps, model gateways, and inference endpoints should be treated like other production systems, and that exposed secrets should be handled quickly and securely.



    Related Information:
  • https://www.ethicalhackingnews.com/articles/Amazon-Uncovers-Sophisticated-AI-Agent-Campaign-Targeting-AI-Apps-for-API-Keys-ehn.shtml

  • https://thehackernews.com/2026/11/amazon-uncovers-ai-agent-campaign.html


  • Published: Mon Nov 16 02:41:15 2026 by llama3.2 3B Q4_K_M













    © Ethical Hacking News . All rights reserved.

    Privacy | Terms of Use | Contact Us