AI safety & security / The AI briefing

Hundreds of ARTEX AI hacking consoles found exposed online, as attacks linked to the tool lean on weak passwords and trusted apps

Censys found more than 1,400 internet-facing consoles of ARTEX, an AI framework that automates penetration testing and has been linked to real attacks. Microsoft says one group used it to attempt sign-ins to more than 2,250 tenants. What small firms and their IT providers should check.

A hand holding a CryptoCard security token showing a one-time passcode in front of a login screen with user name, password and passcode fields
A one-time passcode token held in front of a login screen, 2010. Illustrative photo, not from the incidents described. Photo: Brian Ronald (via Wikimedia Commons), licence CC BY-SA 4.0; cropped; this crop is shared under the same licence.

Security firm Censys says that as of 4 November it saw 1,465 internet-facing hosts running the console of ARTEX, an open-source AI framework that automates penetration testing, or about 815 distinct-looking deployments. Security companies have linked the tool to real attacks, including a breach of ConnectWise customer tenants that Microsoft attributes to a China state-aligned group. TechCrunch, reporting the findings on 6 November, notes that many ARTEX-linked attacks relied on weak or known passwords and missing multi-factor authentication.

What Censys found

ARTEX is an open-source framework, released on GitHub in July 2026, in which AI agents run a penetration test, the authorised probing of systems for weaknesses, from reconnaissance to data theft. In a blog post dated 5 November, Censys says that "as of November 4, 2026, Censys observed 1,465 hosts running an exposed ARTEX console". About 650 sit in two blocks of identically configured hosting, which leaves roughly 815 distinct-looking deployments. Censys notes the console does not need to face the internet and that overall use is likely higher.

Some exposed servers held more than the console. Censys counted 155 ARTEX hosts also running other attack tools it detects, and 32 open directories on ARTEX hosts exposing post-exploitation kits and implants, the hidden programs attackers leave behind to keep access. TechCrunch quotes Censys researcher Silas Cutler as saying not every instance is run maliciously and many are used by authorised testers.

After CrowdStrike published a report on malicious use in October, the developer said they were not involved in the attacks and that the project would go closed-source with no further releases. Censys says the repository has been taken down, but archived copies and rebrands are still circulating.

How attackers used it

In an October report, Microsoft says a China state-aligned actor it tracks as Storm-0940 used ARTEX to automate scanning, password spraying (trying common passwords across many accounts) and abuse of OAuth applications, the third-party apps that a business has already allowed into its Microsoft 365. Microsoft assesses that the group was responsible for the ConnectWise supply chain compromise involving the company's Cloud Backup application.

After that compromise, Microsoft says, the group used ARTEX to attempt sign-ins to more than 2,250 tenants. In one earlier campaign between 10 and 14 October, it recorded more than 151,000 failed logons against 327 accounts at eight organisations, peaking at 2,900 failures per minute; that campaign produced no successful sign-in. Elsewhere, Microsoft says, several ARTEX-driven attacks reached privileged cloud access, in multiple cases less than three hours after targeting began, although "the majority of the targeting wasn't successful."

Microsoft adds that the activity "did not result from a compromise of or a vulnerability in Microsoft corporate systems, products, or services," and that a successful sign-in by the attacker does not necessarily mean customer data was accessed.

The weak point was ordinary

TechCrunch sums up the pattern: "Many of the attacks linked to Artex involved exploiting simple vulnerabilities, such as guessing weak or known passwords, and a lack of multi-factor authentication." Microsoft's write-up also highlights "how a compromised third-party application with high privileges could provide access beyond its originally intended purpose."

What changes with an AI agent is speed and stamina. Microsoft notes that agents "can work around the clock, on weekends, and on holidays" and try several routes at once, which leaves defenders little time to respond to alerts. That matters for small firms, many of which have no one watching sign-in logs at the weekend.

Questions for your IT provider

Most small businesses never see ARTEX directly. They are exposed through their own accounts and through the apps and IT suppliers connected to them. Evoogen suggests asking whoever runs your Microsoft 365 or Google Workspace a short set of questions, based on the weaknesses Microsoft and TechCrunch describe.

Is multi-factor authentication on for every account, including old staff and shared mailboxes? Which third-party apps have access to the tenant, and do any have more permissions than they need? Did you or your provider ever use ConnectWise Cloud Backup, and if so, has the review Microsoft and ConnectWise recommend been done? How long are sign-in and audit logs kept? And who would act if an alert came in at the weekend?

Each answer needs a named owner, and a recheck after staff or suppliers change.

If you want a clear list of the accounts, apps and suppliers that can reach your business data, and the gaps closed, Evoogen can review your setup and fix what it finds. Talk to Evoogen about securing your tools.

Sources

Evoogen checked these sources on 9 November 2026. Company and research claims stay attributed to their source; Evoogen analysis is labelled.

  1. TechCrunch: Hundreds of Artex AI hacking tools found publicly exposed to the web (6 November 2026) — checked 2026-11-09
  2. Censys: ARTEX: Agentic Asbestos (Silas Cutler, 5 November 2026) — checked 2026-11-09
  3. Microsoft Security Blog: AI-driven, large-scale nation-state campaign using ARTEX enables supply chain compromise (21 October 2026) — checked 2026-11-09