The FBI has warned that hackers are using an open-source artificial intelligence hacking framework called ARTEX to break into corporate and government networks, with some attacks progressing from initial targeting to privileged access in under three hours. The tool automates several stages of cyberattacks, including network reconnaissance, vulnerability discovery, exploitation and data theft, reducing the amount of technical work attackers need to perform manually.

In a cybersecurity advisory issued on November 13, 2026, the FBI said ARTEX had been weaponized by multiple threat actors since at least August. The campaigns have targeted financial institutions, government agencies, technology companies, defense organizations, healthcare providers and other critical industries worldwide. The agency warned that the tool allows attackers with limited technical expertise to conduct complex operations against multiple organizations simultaneously.

What Is ARTEX and How Does the AI Hacking Tool Work?

ARTEX is an open-source autonomous penetration-testing framework powered by large language models (LLMs). It was originally developed to help organizations test their cybersecurity defenses and identify weaknesses in their networks.

Unlike conventional hacking tools that require users to execute individual commands and manually coordinate different stages of an attack, ARTEX uses AI agents to automate multiple activities.

Users can provide high-level objectives, which the framework converts into technical tasks performed by specialized agents.

Its capabilities include:

  • Network reconnaissance: Scanning systems and identifying exposed services.
  • Vulnerability discovery: Searching for weaknesses in applications and network infrastructure.
  • Exploitation: Taking advantage of vulnerable systems or inadequate security configurations.
  • Privilege escalation: Attempting to obtain higher levels of access within compromised environments.
  • Lateral movement: Searching for additional systems and resources after gaining an initial foothold.
  • Data exfiltration: Collecting and transferring sensitive information from compromised networks.

The framework can connect to external AI models, including OpenAI’s GPT, Anthropic’s Claude and other open-source or international models.

This means ARTEX does not depend on a single proprietary AI model. Instead, it uses external models to support the reasoning and execution capabilities of its autonomous agents.

FBI Says Hackers Can Compromise Networks in Under Three Hours

The most concerning finding in the FBI advisory is the speed at which ARTEX can automate cyber intrusions.

According to the agency, some attacks have progressed from initial targeting to privileged access in less than three hours.

This is significant because traditional cyberattacks can involve several stages that require attackers to manually identify vulnerabilities, test access methods and navigate compromised networks.

ARTEX reduces the time required to coordinate these activities by allowing AI agents to work through multiple stages with limited human intervention.

The FBI also warned that the framework makes it easier for individuals with relatively limited technical expertise to conduct sophisticated operations.

Although ARTEX-generated activity can produce large volumes of suspicious network traffic, the speed of the attacks can make it difficult for security teams to investigate and respond before attackers gain access.

The problem is particularly serious for organizations that depend on manual security monitoring and incident response.

South Korean Banks Among the Major Targets

One of the most significant ARTEX-related campaigns targeted South Korea’s financial sector.

According to the FBI, attackers used the framework against several South Korean financial institutions, resulting in the theft of personal information belonging to approximately 67,000 people.

The attacks exploited weaknesses in cybersecurity defenses, including inadequate authentication protections and insufficient security controls.

The campaign demonstrated how AI-powered automation could be used to target multiple organizations and move between different stages of an intrusion.

At least nine South Korean banks were reportedly targeted during the campaign, prompting investigations by local authorities.

The incident also highlighted the importance of multi-factor authentication (MFA), which requires users to verify their identities through an additional authentication method.

Without adequate authentication controls, stolen or exposed passwords can provide attackers with an easier route into enterprise systems.

ConnectWise Supply Chain Attack Exposes More Than 2,250 Customers

ARTEX was also used in a cyberattack involving ConnectWise, an American IT management software provider.

The attack targeted ConnectWise Cloud Backup and resulted in unauthorized access to Microsoft 365 environments belonging to more than 2,250 downstream customers.

The incident is particularly important because it demonstrates how a security breach at one technology provider can affect a much larger network of businesses.

ConnectWise provides technology services used by managed service providers (MSPs), which manage IT infrastructure and applications for other organizations.

When attackers compromise a provider or its systems, they may potentially gain access to information belonging to multiple customers.

Microsoft previously identified the campaign and assessed that the attackers were likely linked to China.

The FBI subsequently confirmed the use of ARTEX in the attack, reinforcing concerns about the growing role of autonomous AI tools in supply chain cyberattacks.

Barracuda Also Reports an ARTEX-Linked Security Incident

Cybersecurity company Barracuda also disclosed that it had been affected by an ARTEX-related attack.

According to the company, attackers discovered an exposed credential associated with a third-party service provider whose access had not been revoked after its work with Barracuda ended.

The compromised credential allowed attackers to access stores of credentials that Barracuda used to collect information from customer systems.

The company said it could not rule out unauthorized access to customer environments because its available logs did not provide sufficient visibility.

Barracuda subsequently decided to assume the worst and rotate all customer credentials.

The incident highlights an important security issue: even when a company ends its relationship with a vendor, old credentials and access permissions can remain active if they are not properly revoked.

ARTEX Developer Removes Open-Source Version but Hackers Continue Using It

ARTEX was originally released publicly on GitHub in July 2026 by a developer using the name Autumn-27.

The framework was initially presented as a tool for helping organizations conduct automated penetration testing and improve their security defenses.

However, after cybersecurity researchers identified its use in malicious campaigns, the developer announced that the project would no longer be maintained as an open-source release.

The original GitHub repository was removed in October 2026.

Despite that decision, the FBI warned that ARTEX remains widely accessible through community forks, archived versions, localized variations and a commercial edition called ARTEX PRO.

This means removing the original repository has not eliminated the availability of the framework.

The situation illustrates a wider challenge associated with open-source AI tools: once software becomes publicly available, developers may have limited control over how it is redistributed or used.

Researchers Find More Than 1,600 Exposed ARTEX Systems

Another concern highlighted by the FBI involves the exposure of ARTEX installations on the public internet.

Researchers identified more than 1,600 publicly exposed ARTEX console instances as of early November 2026.

These systems were accessible online, potentially because of incorrect configurations or inadequate access protections.

The actual number of deployments could be significantly higher because the identified systems do not include installations operating behind firewalls or within private networks.

Publicly exposed consoles can create additional opportunities for unauthorized users to access AI-powered hacking capabilities.

The discovery also demonstrates that the same tools used to test cybersecurity defenses can become security risks when deployed without appropriate protections.

Which Industries Are Being Targeted?

The FBI identified several industries affected by ARTEX-enabled cyber campaigns.

IndustryPotential security risks
Financial servicesExposure of customer information and financial systems
GovernmentUnauthorized access to sensitive government networks
TechnologySoftware and cloud infrastructure compromises
DefenseExposure of sensitive operational information
HealthcareAccess to patient and medical information
EnergyDisruption of critical infrastructure
TelecommunicationsCompromise of communications systems
ManufacturingOperational and supply chain disruptions
Legal servicesExposure of confidential client information
SemiconductorTheft of sensitive business and technical information

The wide range of targeted industries indicates that ARTEX is not limited to a particular type of organization.

Any business operating exposed applications, cloud infrastructure, employee portals or third-party integrations could potentially face risks if its security controls are inadequate.

FBI Issues Cybersecurity Recommendations to Businesses

The FBI has encouraged organizations to strengthen their cybersecurity defenses against automated attacks.

The agency’s recommendations include reviewing internet-facing applications, protecting cloud infrastructure and monitoring suspicious activity across enterprise networks.

Important security measures include:

  1. Enable multi-factor authentication: Require additional identity verification, particularly for administrative accounts and remote access.
  2. Secure exposed applications: Review internet-facing services and remove unnecessary access points.
  3. Protect third-party integrations: Regularly examine vendor permissions, API credentials and external connections.
  4. Monitor authentication activity: Investigate unusual login patterns, unexpected privilege changes and suspicious access attempts.
  5. Improve API security: Apply appropriate rate limits and authentication controls to sensitive application interfaces.
  6. Monitor data transfers: Investigate unusual outbound traffic and unexpected access to databases or cloud storage.
  7. Review security logs: Maintain sufficient logging to identify suspicious activity and investigate incidents quickly.

The FBI has also published indicators of compromise and technical details to help organizations identify activity associated with ARTEX.

The Bigger Picture

The ARTEX warning highlights how artificial intelligence is changing the economics and speed of cybercrime. Tasks that previously required considerable technical knowledge and manual coordination can increasingly be automated through AI agents. This does not mean AI independently creates every vulnerability, but it can help attackers identify and exploit existing weaknesses much faster.

The incidents involving South Korean financial institutions, ConnectWise customers and Barracuda demonstrate how these attacks can affect organizations across different industries. The broader lesson for businesses is that basic cybersecurity measures remain essential. Strong authentication, properly managed credentials, secure third-party access and continuous monitoring can reduce the opportunities available to automated hacking systems.

Looking Ahead

The FBI is expected to continue monitoring ARTEX-related campaigns and sharing threat intelligence with organizations and cybersecurity professionals. Although the original developer removed the public GitHub repository, the availability of archived versions and community forks means the framework could remain a threat. Businesses will need to review their security infrastructure and ensure that vulnerabilities are addressed before automated attackers can exploit them.

The growing use of AI-powered hacking tools is also likely to accelerate investment in autonomous cybersecurity defenses. Security companies are developing AI systems capable of identifying suspicious behaviour and responding to threats more quickly. The challenge will be ensuring that defensive AI can match the speed of automated attacks while maintaining accuracy, accountability and human oversight.

Get the day’s top stories in your inbox

One concise email. No spam, unsubscribe anytime.