TechCrunch reported on 6 November that security researchers had found hundreds of copies of a Chinese-made AI hacking tool publicly exposed to the internet for anyone to access. The tool is ARTEX, and the count comes from the internet-scanning firm Censys. As of 4 November 2026, Censys observed 1,465 hosts running an exposed ARTEX console. About 650 of those sit in two identically configured hosting blocks, which leaves roughly 815 distinct-looking deployments.
What ARTEX is
Censys describes ARTEX as an autonomous penetration-testing framework, open source and driven by large language models, that was released in July 2026. TechCrunch describes this kind of tool as software that automatically probes computer systems for weaknesses. Censys says ARTEX was published on GitHub as Autumn-27/ARTEX on 26 July 2026 under the AGPL-3.0 licence. Microsoft has described it as a system of AI agents that can conduct cyberattacks from reconnaissance through data exfiltration with little human interaction.
According to Censys, it is written in Go, uses a Next.js web console backed by PostgreSQL, includes mitmproxy for traffic recording, and works with Anthropic, OpenAI or any OpenAI-compatible model endpoint. Censys says its developer, Autumn-27, also wrote ScopeSentry, a widely used Chinese asset-discovery and vulnerability-scanning platform with more than 2,000 GitHub stars. Censys estimates ARTEX reached 1,500 to 2,000 stars within ten weeks of release. TechCrunch says it is free to download and run on one's own servers, and popular with people who cannot access Anthropic or OpenAI's most advanced cybersecurity AI models.
What the exposed systems reveal
TechCrunch says Censys identified more than 800 distinct deployments accessible from the web without a password, though Censys's own post describes them only as exposed consoles. Censys notes that the ARTEX console does not require public-facing access, and that overall global deployment is likely higher than its count. The gap between the raw host count and the roughly 815 distinct-looking deployments is worth keeping in view when reading the headline figure.
TechCrunch reports that the publicly accessible systems also expose attack tools such as implants used to compromise and maintain persistent access to a victim's network. Censys says 155 ARTEX hosts, or 10.6%, also run at least one other threat it detects, including AI attack frameworks, recon platforms, tunnelling tools and command-and-control software such as Vshell, Cobalt Strike and Adaptix C2. It also counts 32 open directories on ARTEX hosts, exposing post-exploitation kits, stagers and implants.
Two October campaigns
According to Censys, two threat intelligence firms tied ARTEX to attackers' operations in October. In Censys's summary, CrowdStrike connected the tool to a campaign aimed at South Korean financial institutions, while Microsoft tied it to Storm-0940's compromise of tenants belonging to ConnectWise customers.
Censys's summary of CrowdStrike's 7 October 2026 report says an actor CrowdStrike did not attribute used ARTEX against South Korean financial institutions between late September and early October. CrowdStrike assessed with moderate confidence that the actor was a Chinese speaker motivated by money, Censys says. According to that summary, the actor got into a loan inquiry system used by brokers at one bank and a mobile system for staff at another, and took data. Censys's account of the report says the actor's ARTEX instance was hosted at 38.244.50[.]120 and relied mainly on the DeepSeek v4.1-flash model, and that the actor also used Claude Code sessions running on GLM and Grok.
Censys says CrowdStrike found the activity because the actor left open directories exposing a .claude folder, prompts, session histories and ARTEX configuration files. Censys says it saw open directories on that server from 18 August 2026, seven weeks before CrowdStrike's report, exposing exploit binaries and a .claude folder. Censys also says the file names show the actor working on targets well beyond South Korean banks, including payment and gaming services.
Microsoft has blamed Storm-0940 for the supply-chain compromise of ConnectWise Cloud Backup, according to Censys, which describes the group as a China state-aligned actor that has been active since 2021. According to Censys's summary of Microsoft's report, the group used ARTEX to automate scanning, password spraying and the abuse of OAuth applications. Censys relays Microsoft's account that the group attempted sign-ins to more than 2,250 tenants, peaking at 2,900 attempts per minute, and that some victims saw privileged cloud access in under three hours. Microsoft also notes it has seen multiple actors across regions using ARTEX, according to Censys.
The developer's response
Censys says that on 8 October, the day after CrowdStrike's report, Autumn-27 posted a statement on their GitHub profile. In it, the developer said they had no involvement in the attacks and that the tool was intended for testing within an authorised scope of assets. Autumn-27 then stated the project would go closed-source with no further releases. Censys says the repository has since been taken down, but archived copies, translations and rebrands such as ScopeWeaver are still circulating. TechCrunch likewise says there are still dozens of copies of the code on code-sharing sites that can be used in active attacks.
Sources
Whittaker, Z. (2026) 'Hundreds of Artex AI hacking tools found publicly exposed to the web', TechCrunch, 6 November 2026, available at: https://techcrunch.com/2026/11/06/hundreds-of-artex-ai-hacking-tools-found-publicly-exposed-to-the-web/ (Accessed: 6 November 2026).