# FBI Warns ARTEX Is Automating Intrusions; Amazon Says Stolen AI Keys Fund the Agents

Mohit Tyagi · 2026-11-14 · Security

[https://timesof.ai/2026/11/fbi-flash-artex-machine-speed-intrusions-amazon-stolen-ai-keys](https://timesof.ai/2026/11/fbi-flash-artex-machine-speed-intrusions-amazon-stolen-ai-keys)

> The FBI issued a FLASH alert on November 13 saying a variety of threat actors have used ARTEX, an open-source, LLM-driven penetration-testing framework, since at least August 2026 to automate intrusions from reconnaissance to data theft, with some reaching privileged access in under three hours. The bureau says the tool still spreads through forks, archived copies and a commercial version after its developer pulled it from GitHub, and cites research counting more than 1,600 exposed ARTEX consoles. The same day, Amazon said decoy sensors captured ARTEX agents' own campaign records showing Chinese-nexus groups stealing AI model keys to pay for further key hunting; it could not verify the outcomes the agents reported.

The US government has now put its name to what security firms have been reporting for two months: an AI attack framework anyone can download is running real intrusions. The FBI issued a FLASH alert on **November 13** on **ARTEX**, an open-source penetration-testing platform driven by large language models, saying threat actors have "weaponized" it "to automate multiple phases of cyber intrusions at machine speed, from reconnaissance through data exfiltration." On the same day, Amazon published records captured from ARTEX agents themselves, showing a campaign that funded its AI compute with the keys it stole.

## What the FBI says

According to the alert, a variety of malicious actors have used ARTEX since at least **August 2026** against organizations worldwide, in sectors including financial services, government, defense, semiconductors, energy, healthcare and telecommunications. Some intrusions went from first targeting to privileged access in **under three hours**, the FBI says, and the framework lets actors "with minimal technical expertise" run complex operations against several targets at once.

The FBI lists the campaigns it already counts against the tool: financially motivated attacks on South Korean financial institutions that exposed personal information on about **67,000** people, and a supply-chain compromise of ConnectWise Cloud Backup customers that reached more than **2,250** downstream tenants. Times of AI covered both, in [the Korean bank breaches](https://timesof.ai/2026/10/korea-artex-ai-bank-hacks) and [the ConnectWise campaign Microsoft attributed to Storm-0940](https://timesof.ai/2026/10/microsoft-storm-0940-artex-agentic-ai-connectwise-supply-chain).

ARTEX has no model of its own. It plugs into external providers, including frontier models from Anthropic and OpenAI as well as DeepSeek, GLM, Grok and Kimi, the FBI says. Its analysis of attacker-controlled instances found Chinese models most common, with DeepSeek's Flash family dominating "due to their balance of capabilities, relatively low cost, and lack of observable guardrails when framed for penetration testing." That matches what [CrowdStrike reported about the VAPOR PANDA campaign](https://timesof.ai/2026/10/crowdstrike-vapor-panda-artex-deepseek-200-organizations) last month.

Pulling the code did not stop it. The developer took the GitHub repository down on **October 8**, but the FBI says ARTEX continues to spread through community forks, localized versions, archived copies and a commercial edition, ARTEX PRO. It cites research by Censys counting more than **1,600** publicly exposed ARTEX consoles as of early November, with the true number likely higher. The bureau's advice is mostly about exposure: inventory every internet-facing system, including partner portals and legacy services the framework favours, and keep administrative interfaces off the open internet.

## What Amazon saw from inside

Amazon's evidence comes from the attackers' own tooling. Between **October 3 and November 5**, its MadPot decoy sensors posed as LLM inference APIs, and ARTEX agents sent them requests containing their operating memory, CISO **CJ Moses** wrote. Amazon assesses, with moderate confidence, that disparate Chinese-nexus groups were behind the activity.

The largest campaign was a self-sustaining theft of AI credentials: agents searched for exposed AI applications and model gateways, extracted the provider keys they held and fed those keys back into the framework's own gateway to pay for more searching. A second campaign tried to defeat identity checks at two cryptocurrency exchanges with real-time deepfakes; the vendor's server rejected the deepfake in every attempt the agents recorded. Amazon's conclusion is blunt: "None of the techniques in these campaigns were new." What changed is speed and persistence, and the agents succeeded where basic hygiene was missing, such as default credentials and keys left in public code. That is the same prize [Railway said attackers took from its platform](https://timesof.ai/2026/11/railway-breach-stolen-ai-keys-740k-unauthorized-usage).

## Limits

- Amazon says it did not observe the reported activity as it happened and cannot verify the outcomes the agents described; its records show what the actors pursued.
- Both the FBI's and Amazon's attributions rest on technical artefacts and are stated with qualifications; the FBI does not name a single actor behind ARTEX use.
- The FBI's exposed-console count comes from Censys research it cites, not from its own scan, and counts only consoles reachable from the internet.
- The FBI's figures on the Korean and ConnectWise incidents draw on earlier vendor and press reports it footnotes.

## Sources

- [FBI FLASH-20261113-001: Threat Overview of ARTEX Autonomous Penetration Testing Framework (November 13, 2026)](https://www.ic3.gov/CSA/2026/261113.pdf)
- [AWS Security Blog: Amazon Threat Intelligence observes threat actors running cybercrime campaigns with ARTEX AI agents (November 13, 2026)](https://aws.amazon.com/blogs/security/amazon-threat-intelligence-observes-threat-actors-running-cybercrime-campaigns-with-artex-ai-agents/)
- [The Record: FBI flash alert says Artex tool automating cyberattacks at 'machine speed' (November 13, 2026)](https://therecord.media/artex-ai-tool-fbi-flash-alert)