Community Trust ScoreVerified
The FBI issued a Nov. 13 advisory on cyberattacks using the ARTEX AI framework to breach networks and gain privileged access in under three hours. Observed since August, the attacks targeted government, defense, healthcare, and financial services.
More coverage is filed under Technology.
Designed for authorized penetration testing, ARTEX lets operators instruct AI agents using ordinary language. The FBI noted the theft of personal information from approximately 67,000 people in South Korea.
Microsoft and Google track ARTEX activity
A Oct. 21 Microsoft investigation attributed a campaign to Storm-0940, a group aligned with China-state interests. After compromising ConnectWise, the group used stolen credentials to attempt sign-ins to more than 2,250 Microsoft 365 tenants.
Microsoft found reconnaissance and modifications to identities and privileged roles in several environments. It clarified that the 2,250 figure represents attempted access, not confirmed breaches. In another operation, Microsoft recorded 151,000 failed logins during a password-spraying attack against 327 accounts across eight organizations, with no successful authentications.
Google research published Oct. 30 detailed ARTEX activity in Europe and the U.S., affecting telecommunications, semiconductor, and enterprise software firms. Researchers saw activity from espionage groups and financially motivated attackers. ARTEX generated heavy logging and errors; one intrusion produced 240 log events, showing signs of agents interfering with each other.
Crypto and cloud infrastructure targets
Barracuda reported a Nov. 1 intrusion starting with a cloud access key provided to an AI vendor that was not retired. The attacker entered Barracuda’s cloud environment after the vendor’s development environment was compromised in October. Over roughly 44 hours, the intruder accessed around 10 cloud accounts by following credentials.
Barracuda said the credential-hunting routines and traces matched the ARTEX framework, though the tool did not identify the attacker. The intruder searched for seed phrase files, extended private keys, and Ethereum keystore files, but Barracuda confirmed no wallet material existed on the systems. The company stated, “That is on us.”
Amazon reported Nov. 13 that attackers used deepfake video and altered documents to bypass identity verification at two crypto exchanges. Amazon assessed ARTEX involvement with moderate confidence based on records sent to decoy systems between Oct. 3 and Nov. 5. One verification vendor rejected every deepfake attempt, even when the agent completed face-movement sequences.
In a separate campaign, attackers stole credentials for AI services to fund further AI processing. Amazon noted the techniques relied on unpatched software, unauthenticated interfaces, and exposed credentials.
GitBook told customers to replace integration secrets after a Nov. 3 AI-agent attack accessed infrastructure credentials, including Git Sync secrets, API tokens, and GitBook Assistant questions. Everspin detected ARTEX-related attempts in South Korean financial environments. In October, the company announced an AI identity partnership with KB Kookmin Card and SBI Holdings to study delegated permissions, spending limits, and user consent before launching automated payments.
