Barracuda ties 44-hour cloud intrusion to leftover vendor key
Original title: ARTEX AI Framework Linked to 44-Hour Barracuda Cloud Attack Targeting Credentials and Crypto Wallets
Attackers chained stolen cloud secrets into Kubernetes and source code while hunting wallets and reusable credentials.
Barracuda said a 44-hour intrusion across about 10 cloud accounts began with a valid access key issued to an AI vendor and never retired. After an October compromise of that vendor, the attacker took the key from a GitHub configuration file and moved through analytics, source code and Kubernetes, searching for cryptocurrency wallets and secrets. Barracuda links the method to the open-source ARTEX framework, not a named group. Notified XDR customers were told to recreate integration credentials.
Unconfirmed: Logging cannot exclude access to XDR credentials, and the ARTEX match is a method link, not group attribution.
- Activity ran from 04:33 UTC on November 1 to 23:31 UTC on November 2.
- Barracuda disclosed the incident on November 10, 2026.
- An attempt to create an administrative cloud identity was blocked.
- Indicators include i31911, i28324probe, i19ghci.yml and i19ci2.yml.
Impact: Cloud secrets stolen; XDR integration credentials not ruled out
- 1.Notified Managed XDR Cloud and Monitored Endpoint customers should recreate integration credentials.
- 2.Review connected-system activity from November 1.
- Vendors
- Barracuda
- Products
- Barracuda XDRManaged XDR Cloud
- Malware
- ARTEX
- Victims
- Barracuda
From the original article
2 min read at gbhackers.com
Barracuda has linked a 44-hour intrusion across approximately 10 cloud accounts to activity consistent with ARTEX, an open-source, AI-driven penetration-testing framework. The attacker chained stolen credentials across cloud infrastructure, analytics, source-code repositories and Kubernetes, while searching for cryptocurrency wallets and reusable secrets. The company disclosed the incident on November 10, 2026.
By Mayura Kathir · gbhackers.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.