The FBI says hackers are using ARTEX, an open-source AI framework, to go from targeting an organization to holding privileged access in under three hours in some intrusions.
The warning, issued in a November 13, 2026 advisory, covers attacks on financial, government, healthcare, defense and technology targets since at least August 2026, including a South Korean breach that exposed data on about 67,000 people.
Microsoft links ARTEX to connectWise supply chain attack
ARTEX AI has also been linked to a major supply-chain compromise involving ConnectWise, a technology provider whose cloud applications serve multiple organizations.
In an October 21 investigation, Microsoft attributed the campaign to Storm-0940, a China-state-aligned threat actor that used compromised application credentials to target hundreds of organizations.
Following the ConnectWise incident, attackers attempted to authenticate against more than 2,250 Microsoft 365 customer tenants using stolen application credentials. Microsoft observed subsequent reconnaissance and attempts to modify application credentials, identities and privileged roles in several customer environments.
However, the figure represents attempted access, not 2,250 confirmed successful breaches.
Microsoft also documented a separate password-spraying campaign targeting 327 accounts across eight organizations. The operation generated more than 151,000 failed login attempts without a successful authentication.
Google’s Threat Intelligence Group and Mandiant independently reported observing malicious actors using autonomous penetration-testing frameworks to conduct reconnaissance, lateral movement and privilege escalation. Researchers also identified repeated execution errors and extensive logging, showing that automated attacks can still leave detectable traces.
The findings suggest that ARTEX AI is not necessarily creating entirely new hacking techniques. Instead, it is combining existing attack methods into faster, coordinated operations that leave security teams with less time to respond.
Crypto wallet credentials and exchange verification become targets
The cryptocurrency industry faces additional exposure because attackers are increasingly searching for wallet credentials and attempting to manipulate identity verification systems.
In a November 11 incident report, cybersecurity company Barracuda disclosed that an attacker entered parts of its cloud environment using an access key previously provided to an artificial intelligence vendor. The company had failed to revoke the credential after the vendor completed its work.
After compromising the vendor’s development environment, the attacker apparently recovered the key and used it to access Barracuda’s infrastructure.
Over approximately 44 hours, the intruder reached around 10 cloud accounts, following exposed credentials between systems. Investigators identified activity consistent with ARTEX, including automated searches for Ethereum keystore files, extended private keys and cryptocurrency seed phrases.
Barracuda confirmed that the compromised servers contained no wallet material. However, the incident exposed how stolen infrastructure credentials can become an entry point for cryptocurrency-related attacks.
Acknowledging the security oversight, the company stated: “That is on us.”
Amazon also reported attempts to bypass know-your-customer (KYC) verification at two cryptocurrency exchanges using deepfake videos and manipulated identity documents.
Its November 13 investigation examined records collected by MadPot, Amazon’s network of decoy threat sensors, between October 3 and November 5.
Although the attackers managed to reproduce facial movements required during one verification process, the system’s server-side checks rejected every recorded deepfake attempt.
Amazon assessed the use of ARTEX in these campaigns with moderate confidence, while cautioning that it could not independently verify every activity described in the agents’ records.
Security experts urge stronger credential protection
The FBI’s warning highlights the growing importance of securing third-party integrations, cloud credentials and internet-facing applications.
The bureau recommends that organizations review authentication records, investigate unusual application activity, remove unnecessary permissions and rotate potentially exposed credentials.
Google’s security researchers have also emphasized established defensive measures, including short-lived credentials, network segmentation, least-privilege access and behavioral monitoring.
The broader concern is that ARTEX AI can exploit weaknesses that organizations may have overlooked for months. An exposed API key, outdated application or poorly configured supplier connection could provide an automated attacker with a pathway into more valuable systems.
The original ARTEX developer reportedly removed the public GitHub repository in October following reports of malicious use. However, the FBI says alternative copies, community forks and commercial versions remain available.
For cryptocurrency businesses, the incidents reinforce the need to protect private keys, isolate wallet infrastructure, strengthen identity verification and monitor unusual access to customer accounts.
The FBI’s advisory makes clear that artificial intelligence has become a significant force multiplier for cybercriminals. As autonomous agents become more capable, organizations may have to respond to attacks that progress faster than traditional security procedures were designed to handle.