Home / Blog / FBI says Artex AI open-source hacking tool used to breach…
Tech News

FBI says Artex AI open-source hacking tool used to breach networks

The FBI warned that hackers used the open-source Artex AI tool to breach ConnectWise and compromise 2,250 customer cloud environments in under three hours.

By Dillip Chowdary • Nov 16, 2026 • Source: TechCrunch

FBI says Artex AI open-source hacking tool used to breach networks
Image: TechCrunch

At a glance

Tool category
Autonomous AI penetration testing platform
Initial release
July public debut
Compromise speed
Under three hours from initial targeting
Affected tenants
2,250 ConnectWise Microsoft 365 customers
Stolen records
68,000 South Korean banking customers
Distribution
Closed-source with circulating legacy versions

The Federal Bureau of Investigation issued a flash cybersecurity advisory warning that threat actors are deploying autonomous artificial intelligence hacking software to compromise networks across the United States and globally. According to TechCrunch's report, malicious operators have used an open-source penetration testing tool called Artex to breach financial services, government bodies, technology providers, defense contractors, telecommunications firms, and law practices. Federal investigators warned that these automated campaigns can advance from initial reconnaissance to complete system compromise in under three hours.

This analysis examines how automated intrusion agents pair with commercial large language models, exploit basic credential hygiene flaws, and breach multi-tenant cloud architectures. The reporting is compiled for enterprise defenders, systems engineers, and compliance officers who must secure corporate infrastructure against machine-speed offensive tooling that outpaces conventional human security responses.

01Frontier model integration and three-hour breach speeds

Artex is an open-source offensive security platform designed to evaluate network defenses by simulating real-world intrusion workflows. Malicious operators integrate the software with hosted frontier large language models from Anthropic and OpenAI, as well as foreign models and unhosted open-source machine learning weights. These neural networks process live telemetry from scanned networks, allowing the software to determine offensive paths and make automated execution decisions.

Federal investigators noted that intrusions launched through Artex generate substantial network traffic and data volume that can trigger internal defensive alarms. However, the automated velocity of these operations prevents human defenders from analyzing telemetry and deploying countermeasures before systems fall. In multiple confirmed incidents, threat actors progressed from identifying an exposed perimeter to achieving active administrative compromise in under three hours.

02Autonomous reconnaissance and lateral network movement

Offensive operations typically commence when the platform performs automated scans against an enterprise perimeter to detect accessible attack surfaces. Artex actively searches for common security flaws, including missing multi-factor authentication, unpatched services, and leaked or stolen login credentials. Once it identifies an exploitable flaw, the software automatically executes intrusion routines to secure an initial administrative foothold inside the victim's infrastructure.

After securing initial internal access, the tool begins autonomous lateral traversal to locate and exfiltrate high-value assets across corporate networks. The software enumerates interconnected servers to harvest additional administrative tokens, inspect databases, and access employee mailboxes containing sensitive corporate correspondence. This programmatic expansion allows threat groups to map organizational topologies and gather intelligence without requiring manual intervention from human operators.

Because Artex targets basic operational misconfigurations rather than novel exploits, perimeter hygiene serves as the primary line of defense. Organizations lacking centralized credential governance and strict identity verification provide optimal environments for automated traversal. When defensive perimeters lack robust access enforcement, automated agents can pivot across entire subnets within minutes of the initial penetration.

Advertisement

Tech Pulse Daily

Get tomorrow's pulse first

Join engineers who read Tech Pulse before stand-up. Free, weekday mornings.

03Compromised ConnectWise tenants and South Korean banking records

Federal authorities reported that threat actors have actively deployed Artex across global targets since the software debuted publicly in July. Attackers used the framework to breach multiple financial institutions in South Korea, exploiting weak security defenses and the absence of multi-factor authentication controls. Those financial intrusions culminated in the unauthorized exfiltration of personal records belonging to at least 68,000 individuals.

The FBI also validated findings published by Microsoft indicating that suspected Chinese threat actors deployed Artex against American IT management giant ConnectWise. The hackers breached internal systems and gained unauthorized access to the Microsoft 365 cloud environments of at least 2,250 ConnectWise corporate customers. Representatives for ConnectWise did not respond to requests for comment regarding the federal advisory or the customer impact.

04Unrevoked vendor access and Barracuda credential resets

Cybersecurity company Barracuda separately confirmed that it sustained an intrusion linked to Artex operations conducted by financially motivated cybercriminals. The automated tool discovered an exposed credential tied to a third-party vendor service that Barracuda had failed to revoke after concluding business relations. That unrevoked credential allowed the attackers to access internal credential repositories used by Barracuda to collect operational data from customer systems.

Because existing system logs lacked sufficient detail to rule out broader unauthorized access, Barracuda elected to assume the worst and rotated all customer credentials. Barracuda chief information security officer Arve Kjoelen cautioned that defenders who rely on detecting anomalous behavior before initiating responses will routinely arrive after attackers have completed their objectives. Barracuda did not respond to requests for comment regarding the number of impacted customers.

05Circulating repository code and exposed internet deployments

Following widespread exploitation in cyberattack campaigns, the original developer converted Artex into a closed-source commercial product to prevent further malicious adoption. Nevertheless, federal investigators confirmed that earlier open-source codebases continue to circulate and remain widely accessible across developer communities. These legacy repositories provide threat actors with fully functional autonomous scanning and exploitation capabilities that operate independently of the developer's current controls.

Security researchers independently discovered hundreds of active Artex installations exposed to the public internet without password protections, likely due to operational misconfigurations. These unsecured deployments included the specific Artex system that attackers used to execute intrusions against South Korean financial institutions. Unprotected installations allow outside actors to hijack powerful offensive capabilities and launch autonomous campaigns without hosting their own infrastructure.

Note: Although the primary developer shifted Artex to a closed-source model, federal authorities confirmed that legacy open-source distributions remain widely accessible online. Enterprise security teams must treat autonomous offensive tools as an active threat and enforce multi-factor authentication, rigorous third-party credential revocation, and continuous surface monitoring.

Frequently asked questions

What is Artex and how does it operate?

Artex is an AI-powered penetration testing platform that scans networks for vulnerabilities and weak credentials. Threat actors connect it to large language models from Anthropic and OpenAI to autonomously navigate and compromise corporate systems.

How quickly can an Artex attack compromise a network?

According to the FBI, attacks leveraging Artex can progress from initial network targeting to active administrative compromise in under three hours.

Which organizations were impacted by Artex intrusions?

Hackers used Artex to breach IT management company ConnectWise, accessing the Microsoft 365 cloud environments of at least 2,250 customers. Attackers also stole personal data from 68,000 people across South Korean financial institutions and breached cybersecurity firm Barracuda through an unrevoked third-party credential.

Is Artex still accessible to cybercriminals?

Although the developer made Artex closed-source to deter misuse, the FBI confirmed that earlier open-source versions continue to circulate widely online. In addition, researchers discovered hundreds of misconfigured, passwordless Artex systems accessible across the public internet.

Sources

Dillip Chowdary

Author

Dillip Chowdary

Writes Tech Bytes coverage of AI, engineering, and the tools that actually ship. Editor of Tech Pulse Daily.

Related on Tech Bytes

Advertisement

5-min tech signal

Weekday briefing for engineers who skip the noise.

No spam · Unsubscribe anytime

Advertisement

✈️ CareerPilot

Your AI job-search copilot

Match your resume against live Ashby, Greenhouse & Lever openings — fit scores, job-specific resume optimization and email alerts.

Find matching jobs →

Free Tools

Browse all tools →