Skip to content
NewA calmer, faster ZeroHour
StoryThreat actorhigh

Amazon links ARTEX agents to LLM key hunts and failed deepfake KYC

3 sources · 3 articlesFirst reported Updated ()1
What's new

November 16: Security Affairs adds that Amazon notified affected AWS customers, published IPs seen October 14–November 2, and describes a multi-agent knowledge graph plus an 11.5 fps deepfake pipeline.

Merged summary · x-ai/grok-4.7 · rewritten as coverage arrivesSummaries by Vani AI

Amazon has only moderate confidence China-linked operators used ARTEX from October 3 to November 5 to hunt LLM keys; thefts remain unverified.

Amazon Threat Intelligence says financially motivated operators used the open-source ARTEX agent framework between October 3 and November 5, 2026, to search internet-exposed AI apps and gateways for stored provider API keys. MadPot honeypots posing as LLM APIs captured prompts, tool histories, and campaign records showing stolen keys reused to look for more keys. Amazon has moderate confidence the agents ran on ARTEX under China-linked operators and, according to The Hacker News, cannot verify the reported thefts. Named targets include LiteLLM, One API, New API, Dify, Langflow, RAGFlow, Open WebUI, and Ollama; the playbook cites Langflow CVE-2025-3248 and RAGFlow CVE-2026-28797. Documented real-time deepfake attempts against KYC checks at two cryptocurrency exchanges failed, and Amazon notified AWS customers it believed were affected.

Key facts
  • Observation window was October 3 through November 5, 2026.
  • Amazon rates ARTEX use and China links at moderate confidence.
  • The Hacker News says Amazon cannot verify the reported key thefts.
  • Playbook cites Langflow CVE-2025-3248 and RAGFlow CVE-2026-28797.
  • Amazon published IPs observed from October 14 to November 2.
  • Every documented deepfake KYC attempt failed server-side checks.
Analyze this story with Vani AIImpact, who is affected and what defenders should do, in a new chat

Timeline

dated events in the reporting · oldest first
  1. Oct 3, 2026MadPot begins capturing ARTEX agent campaign traffic · GBHackers
  2. Oct 14, 2026Crypto-fraud infrastructure address observed; IP publication window starts · GBHackers
  3. Oct 17, 2026Credential-theft address seen through October 19 · GBHackers
  4. Nov 2, 2026End of IP range Amazon later published · Security Affairs
  5. Nov 5, 2026MadPot observation of the agent campaigns ends · Security Affairs

What each outlet adds

beyond the shared facts
  • GBHackersSpecific IPs, October 17–19 activity, and a seven-pixel miss
  • The Hacker NewsUnverified thefts, Cloudflare CI/CD token, and Microsoft GitHub hunting
  • Security AffairsCustomer notices, published IPs, agent roles, and 11.5 fps deepfake

Coverage

oldest first · one row per article · UTC
  1. · 3d ago GBHackersFirst report

    Stolen LLM API keys used to fund ARTEX credential-theft agents

    Amazon’s MadPot honeypots captured ARTEX agent memory from October 3 to November 5; a separate deepfake KYC effort failed.

  2. · 1d ago The Hacker News

    Amazon finds AI agents hunting LiteLLM and Dify for API keys

    Amazon cannot verify the reported thefts and has only moderate confidence the agents ran on ARTEX under China-linked operators.

  3. · 1d ago Security Affairs

    Amazon: ARTEX agents stole AI keys and tested deepfake KYC

    MadPot saw agent traffic from October 3 to November 5; Amazon has only moderate confidence goals were met.

Vulnerabilities in this story

All
CVEVulnerabilityCVSSEPSSSignalsPublished
CVE-2025-3248
Unauthenticated RCE in Langflow /api/v1/validate/code
9.8100%KEVRansomwarePoC ×4
KEV
CVE-2026-28797
RAGFlow: In versions 0.24.0 and prior, a Server-Side Template Injection (SSTI) vulnerability exists in RAGFlow's Agent workflow Text…
8.7<1%PoC