FBI Says AI Hacking Tool Artex Breaks In Within Hours

âš¡ TL;DR
An FBI flash advisory says attackers have used Artex, an open-source AI penetration testing framework, since at least August 2026, sometimes reaching privileged access in under three hours. The tool was linked to a breach exposing data of about 67,000 people at South Korean financial institutions and a ConnectWise supply chain attack. The FBI urges organizations to shrink exposed systems, tighten access and keep logs.

The FBI says an open-source AI hacking tool called Artex is already being used to break into organizations worldwide, with some intrusions going from first contact to privileged access in under three hours. The bureau laid out the threat in a FLASH advisory dated November 13, 2026, numbered FLASH-20261113-001 and marked for unrestricted sharing.

AI hacking tool

The warning is not about a future risk. According to the FBI advisory, “a variety of malicious cyber actors” have used the framework since at least August 2026 against financial services, government, technology, defense, semiconductor, energy, healthcare, manufacturing, telecommunications and legal organizations.

What the Artex AI Hacking Tool Does

The FBI describes Artex, which it writes as ARTEX, as an autonomous, multi-agent penetration testing platform written in Go. A user types high-level goals in plain language, and AI agents turn them into machine commands, “requiring minimal human interaction beyond initial objective setting,” the advisory says.

Artex has no AI model of its own. It plugs into outside providers, and the FBI lists frontier models such as Anthropic’s Claude and OpenAI’s GPT alongside open-source or Chinese-origin models including DeepSeek, GLM, Grok and Kimi. In instances controlled by attackers, Chinese models are the most common, with DeepSeek Flash models dominating because of their capabilities, low cost and “lack of observable guardrails when framed for penetration testing,” the bureau wrote.

The FBI says the tool automates “multiple phases of cyber intrusions at machine speed, from reconnaissance through data exfiltration.” The advisory’s list of observed activity includes fuzzing web directories with more than 500 requests, exploiting SQL injection and unsigned SAML assertions, planting web shells, enumerating more than 110,000 Active Directory objects, checking which Windows machines lacked CrowdStrike Falcon, and creating new 100GB storage volumes to prepare stolen data for removal.

Speed Is the Main Threat

The bureau said the framework lets attackers “with minimal technical expertise” run complex operations against several targets at once. It singled out Artex’s skill at finding weak side doors such as partner and supplier portals, mobile backends, employee apps and legacy services that get less security attention than a company’s main products.

Artex traffic is “forensically noisy” and may set off alerts, the FBI noted, but the attack speed leaves defenders little time to respond before the intruders finish. Barracuda’s chief information security officer, Arve Kjoelen, put it this way in remarks reported by TechCrunch:

“Defenders who rely on spotting malicious activity and then responding to it will often arrive after the attacker has finished.”

TechCrunch reported that Barracuda itself was breached in an Artex-linked attack after the tool found an exposed credential for a third-party service the company had not revoked.

ConnectWise Cloud Backup and South Korean Banks

The advisory names two major campaigns. One hit South Korean financial institutions and exposed personal information of about 67,000 people, according to the FBI; TechCrunch put the figure at at least 68,000. The other was a supply chain compromise of ConnectWise Cloud Backup customers that led to attempted sign-ins to more than 2,250 downstream tenants. Microsoft says the hackers behind that attack are likely Chinese, TechCrunch reported, and a ConnectWise spokesperson said the company does not have the visibility to confirm the number of affected environments.

Amazon has also studied the tool. Its chief information security officer, CJ Moses, said frameworks like Artex “have progressed from experimental use to sustained, multi-objective offensive operations that run with minimal human intervention,” The Record reported. AI misuse in real attacks has surfaced before: in September, The Wall Street Journal reported that Google’s Gemini AI was manipulated to breach three companies.

Pulled From GitHub, Still Circulating

A developer using the name “Autumn-27” released the AI hacking tool on GitHub in July 2026, and it gained an estimated 1,500 to 2,000 stars within 10 weeks, the FBI said. The developer said it was meant for authorized security testing and made the repository unavailable on October 8 after reports of abuse.

That has not stopped it. The FBI says Artex keeps spreading through community forks, localized variants, archived copies and a commercial version called ARTEX PRO. As of early November, researchers had found more than 1,600 publicly exposed Artex consoles, and the bureau said the real number of deployments is likely “significantly higher.”

FBI Steps to Blunt the AI Hacking Tool

The bureau urged organizations, especially those running internet-facing apps, self-service portals, cloud systems and partner or supplier platforms, to act now. Its recommendations include:

  • Shrink the attack surface: inventory every internet-facing application, including forgotten peripheral systems, and restrict cloud admin consoles, backup platforms, hypervisors and other sensitive services to trusted networks.
  • Harden identity: apply least privilege to app permissions, watch for new credentials and privilege grants, and review service-principal sign-ins, rotating credentials if anything suspicious turns up.
  • Protect applications: put rate limits and transaction thresholds on high-impact APIs, alert on sequential record lookups in portals, hide detailed error messages from outsiders, and scan code and pipelines for embedded secrets.
  • Preserve evidence: keep sign-in and audit logs, and for ConnectWise-linked activity review records back to at least October 1, 2026, without delaying urgent containment.

The FBI also asked organizations to watch for unexpected outbound calls to AI services such as api.anthropic.com and api.deepseek.com alongside other suspicious activity. Victims can report incidents to a local FBI field office or the Internet Crime Complaint Center at ic3.gov.

0
Show Comments (0) Hide Comments (0)
0 0 votes
Article Rating
Subscribe
Notify of
guest
0 Comments
Oldest
Newest Most Voted
0
Would love your thoughts, please comment.x
()
x