The FBI has issued a flash advisory warning that threat actors are leveraging Artex, an open-source AI-powered penetration testing platform, to conduct autonomous cyberattacks against financial services, government, technology, defense, telecommunications, and legal sectors in the U.S. and globally. Artex integrates with frontier LLMs from Anthropic and OpenAI, as well as open-source and non-U.S. models, enabling rapid network scanning, exploitation of weak protections and exposed or stolen credentials, and lateral movement to access databases and mailboxes. The FBI reports intrusions can progress from initial targeting to active compromise in under three hours, generating noisy traffic that may trigger alerts but outpaces human response. Since its July public debut, Artex has been used in multiple campaigns. Microsoft attributed the breach of IT management firm ConnectWise to likely Chinese hackers using Artex to access Microsoft 365 environments of at least 2,250 customers. In South Korea, Artex compromised financial institutions, exfiltrating personal data of at least 68,000 individuals, facilitated by weak defenses and absent multi-factor authentication. Barracuda disclosed a breach by financially motivated criminals who exploited an unrevoked third-party credential via Artex to access credential stores for customer data collection; logs were insufficient to confirm scope, prompting a full credential rotation for affected XDR customers. Although Artex’s developer subsequently closed the source, the FBI notes the prior open-source version remains widely accessible. Researchers recently discovered hundreds of publicly exposed, unauthenticated Artex instances online, including the system used against South Korea, due to misconfiguration.
Read original
hackernews