Cybersecurity researchers have uncovered an ongoing campaign that uses artificial intelligence to target WordPress websites, e-commerce platforms, and critical infrastructure in Ukraine. The operation combines common hacking tools, known software vulnerabilities, and AI agents to automate attacks and gain access to compromised systems.
According to a report from cybersecurity firm SOCRadar, the campaign is linked to a Russian-speaking, financially motivated threat actor known as CyberXero, that has reportedly been active since at least July 2026. The operation involves automated attacks against WordPress and e-commerce websites worldwide, as well as targeted attacks against Ukrainian energy and utility organizations.
“CyberXero is the underground-forum alias of a financially motivated Initial Access Broker (IAB), an attribution established during the analysis. An IAB compromises organizations and sells that access to other criminals rather than monetizing it directly. What sets CyberXero apart from a typical IAB is architectural: the actor wraps its toolset in an AI layer that lets a single operator run campaigns at a scale that would normally require a team,” the researchers noted in the report.
The campaign exploits known vulnerabilities, including CVE-2026-63030 and CVE-2026-60137, associated with the wp2shell pre-authentication exploit chain, as well as CVE-2026-4815 (an SQL injection vulnerability affecting the Support Board WordPress plugin).
SOCRadar discovered the campaign after identifying an exposed directory on a server, which contained files from the attacker's Linux workstation, including AI session logs, scripts with plaintext authentication tokens, exploitation tools, reconnaissance results, AI agent configurations, and stolen data. The researchers found that one automated scan targeted 4,708 systems and identified 429 WordPress administration panels that the attacker could access. The operation then deployed 32 web shells within 61 seconds.
The campaign has stolen data from four Ukrainian organizations, with more than 628,000 records linked to Ukrainian individuals. Other confirmed compromises affected organizations in Poland, China, and the Netherlands. The researchers also detected attacks against exposed Redis servers that led to the compromise of two Chinese entities.
CyberXero reportedly uses Anthropic's Claude Code with a group of 51 specialized AI agents to support scanning, reconnaissance, vulnerability exploitation, and post-exploitation activities.
A second AI-based setup connects open-source autonomous penetration-testing agent PentAGI to the Cobalt Strike testing frameworkthrough an AI provider API. Cobalt Strike has previously been observed being abused by attackers for command execution, remote control, and post-compromise operations.
SOCRadar also found evidence that CyberXero attempted to bypass AI safety restrictions by changing session context and restarting conversations. In some cases, the attacker reportedly instructed the AI agent to treat the activity as an authorized penetration test. When a model refused a request, the attacker opened a new session and tried again with a different authorization statement.
However, the session logs also recorded several refusals. Across multiple attempts, the AI model declined requests to install a backdoor, disable a firewall, move laterally through a network, and deploy a web shell. In another case, it repeatedly refused a scan and rejected a forged ownership token.
The targeted Ukrainian infrastructure included seven energy and utilities entities under curated reconnaissance, including the national transmission system operator and the country’s largest private energy holding.