123 Reg Says AI-Driven Attack Hit Customer Data, Points to China

UK registrar 123 Reg told customers an AI-driven attack exposed contact and product data, and suspects a China-nexus group tied to ARTEX intrusions in Korea.
Table of Contents
    Add a header to begin generating the table of contents

    The UK domain registrar 123 Reg has told customers that an intruder accessed their contact and product information through an API, and describes the incident as the first known AI-driven attack against the company. It suspects a China-nexus group, but says the attribution is its own assessment and has not been confirmed.

    What the 123 Reg Notification Emails Say Was Exposed

    Notification emails went out this week. They state that contact information, relationship data, company VAT numbers and product information were exposed from an API. A company spokesperson said this means names, phone numbers, countries, language and currency preferences, account creation dates, the products customers hold and their marketing preferences.

    The intrusion took place on September 18. The company has not said how many customers are affected, and its investigation is ongoing.

    Infrastructure and AI Skills That Led 123 Reg to Suspect China

    123 Reg bases its China-nexus assessment on two findings. The core attacker infrastructure was hosted in China and Singapore. In addition, AI attack “skills” were found on Chinese servers, including copies of skills that Anthropic open sourced and other exploitation frameworks.

    The Suspected Link to Claude Code Abuse Against 30 Organizations

    The company suspects the Chinese state-sponsored group known for abusing Claude Code against at least 30 organizations. That group’s activity has been publicly documented separately, and 123 Reg’s view is that the tooling and infrastructure match it.

    ARTEX and the South Korean Financial Institution Intrusions

    123 Reg also raises the possibility that the actor is the one that used ARTEX, an open-source Chinese agentic penetration tool, against South Korean financial institutions. CrowdStrike published an analysis of that activity on October 7 under reference CSA-261162, describing activity from late September. The registrar’s intrusion on September 18 predates the activity CrowdStrike describes, so the connection rests on shared tooling and not on a matching timeline.

    How Much Weight the AI-Driven Label Deserves

    The registrar’s description of an “AI-driven” attack rests on the presence of AI tooling on the attacker’s servers, not on a published account of how the models operated during the breach. The distinction matters. Finding agent skills and frameworks on infrastructure shows that the operator had AI-assisted capabilities available; it does not by itself show what share of the intrusion those tools carried out.

    The stolen data is limited in type: contact details and product information, with no mention of payment data or passwords. Even so, a registrar’s customer list is useful to an attacker. Names, phone numbers, the domains and services a business holds, and VAT numbers could support targeted phishing against customers who manage websites and email.

    The case adds to a small but growing set of incidents in which defenders point to agentic tooling in attacker infrastructure. Because attribution here is the company’s assessment, the items to watch are whether the security firms that tracked the earlier campaigns confirm the link and whether 123 Reg publishes a customer count. Neither has happened yet, and the registrar has said only that the investigation continues.

    The registrar’s disclosure is notable for how it describes the evidence. Instead of reporting only that an intruder reached an API, 123 Reg chose to list what it found on the attacker’s servers, including open-sourced Anthropic skills. The company names China and Singapore as the hosting locations for core infrastructure. A UK business that discloses attacker tooling this openly gives other registrars and hosting providers indicators they can check against their own logs.

    The registrar’s customers are mostly small businesses and individuals who register domains and run websites, and the notification emails are the only guidance the company has issued so far. No customer count, no description of how the API was reached and no timeline for completing the investigation accompany the disclosure, so the full effect on customers remains unmeasured.

    Related Posts