Attackers are exploiting CVE-2026-82329, a critical authentication bypass in JFrog Artifactory. Attackers can grant themselves admin privileges. watchTowr observed the first exploitation on internet-exposed systems within days of the patch being released. Artifactory was already prominently exploited this summer when it served as an attack vector against Hugging Face via an OpenAI model.
JFrog disclosed the vulnerability last Friday. On Tuesday, watchTowr’s threat intelligence team reported that attackers were already actively targeting servers exposed to the internet. The vulnerability has a CVSS score of 9.8 and falls under CWE-287, improper authentication. No authentication, no privileges, and no user interaction are required for exploitation.
This affects self-hosted installations in their default configuration. According to JFrog’s advisory, fixes are available in versions 7.111.21, 7.117.28, 7.125.20, 7.133.29, 7.146.38, and 7.161.20, depending on the release branch.
Admin tokens and reconnaissance
watchTowr observed in its honeypot network how attackers were generating admin tokens for themselves. In addition, users, groups, credential sets, and federated access configurations were mapped. So far, this involves a small number of IP addresses originating from various regions. There is no evidence of large-scale scanning yet, but watchTowr does not expect this to remain the case.
The security firm therefore advises organizations running vulnerable versions to patch Artifactory immediately and treat their systems as if they have been compromised. It is therefore important to review audit logs, rotate credentials, and check connected systems for backdoors.
Not the first incident
Artifactory is used to manage software artifacts, packages, binaries, and AI models. The platform was already in the news prominently this summer. In July, OpenAI and JFrog reported that an AI agent exploited a zero-day vulnerability in self-hosted Artifactory to break out of its sandbox and reach Hugging Face’s production environment. That vulnerability was fixed in version 7.161.
It is unclear whether humans or automated agents are behind the current attacks. In any case, organizations often lack sufficient visibility into their own supply chains.