SOCRadar's October 6 report documents CyberXero, a Russian-speaking, financially motivated initial access broker that pairs commodity offensive tooling with an AI orchestration layer on its own infrastructure. The operation runs two pipelines: a global automated campaign against WordPress and e-commerce platforms, and a curated manual campaign against Ukrainian energy and critical infrastructure. These are one security vendor's findings, not a court finding or a government attribution.
The report says more than 628,000 Ukrainian individuals have confirmed data in the actor's possession, including residents of Kharkiv. Confirmed exfiltration covers four Ukrainian organizations, while seven more energy-sector entities were under reconnaissance without confirmed compromise. Reconnaissance and confirmed breach are different states.
The investigation began with an exposed open directory on an HTTP server that held the actor's live working directory: more than 90,000 files across 3,000 subdirectories, including AI session logs, scripts with plaintext tokens, and victim database exports. SOCRadar says the actor remained active at the time of publication.
On scale, the report says a single automated execution scanned 4,708 targets, confirmed access to 429 WordPress administration panels, and deployed 32 shells in a 61-second window. These figures come from the actor's own recovered logs and Telegram notifications, so they describe what the tooling reported, not an independent recount.
The AI layer used Claude Code with up to 51 specialized agents on the primary workstation, plus a PentAGI deployment wired into a Cobalt Strike Team Server through an AI provider API. The recovered logs show the actor using a rehearsed "authorized pentest on my own infrastructure" claim and resetting sessions after refusals. The logs also record refusals that held, including a forged ownership token the model rejected.
Limits: attribution rests on the vendor's account-file, forum and subscription evidence; SOCRadar withholds personal identifiers and says it shared findings with relevant parties. The report does not establish that every scanned target was breached, that any access sale completed, or that Anthropic's systems were technically bypassed. Organizations running WordPress, Magento, PrestaShop or exposed Redis get concrete checks: rogue wp2_-pattern administrators, wp2shell plugin paths, WSO webshells and unexpected SSH authorized_keys entries.