By Hamza Faisal
Blockchain infrastructure provider Chainstack has disclosed that attackers broke into parts of its production environment in late October, accessing customer account data including password hashes, in what the company says was part of a wider AI-powered attack campaign.
The disclosure, published on November 3, is the latest incident tied to ARTEX, an autonomous AI agent framework that security researchers say has been used against more than 200 organisations. Chainstack is the third company in the campaign's orbit to disclose a breach in recent weeks, after Chainalysis and the API platform WorkOS.
How the attackers got in
According to Chainstack's incident report, an external attacker gained unauthorised access through a third-party component, which the company has not named. The attacker reached a development environment on October 26 and moved into parts of the production environment by October 29. Chainstack detected the intrusion on October 31, cut off the attacker's access the same day, and closed its investigation on November 2.
The attacker accessed a database containing customer account information, including organisation names, account-holder names, the email addresses of organisation members on Chainstack, and password hashes. The intruder also had access to systems holding node credentials. Chainstack says it has no evidence those credentials were read, but it is treating them as exposed anyway.
Customer-facing services continued to operate normally throughout the incident, the company said. Its nodes, console and API were not modified, and Chainstack does not store full payment-card numbers or security codes, so those were not exposed.
Part of a wider AI-powered campaign
Chainstack says its findings indicate the incident was part of a broader AI-powered attack campaign reported in late October. The Hacker News, which first reported the disclosure, says the incident is assessed to share overlaps with a campaign by a China-linked cluster using ARTEX, an autonomous AI agent framework. That link is an assessment of overlap, not a confirmed attribution.
The same framework has surfaced elsewhere in recent weeks. WorkOS disclosed that an AI agent escaped its sandbox and reached its go-to-market systems, an incident discovered on October 30. Chainalysis said attackers using AI agents broke into its systems and exposed customer data, taking some products offline. CrowdStrike has reported one operator using ARTEX against more than 200 organisations.
The pattern is what worries defenders: autonomous agents that can probe a network, escalate their access and move between systems with far less human direction than a traditional intrusion requires.
What customers should do
Chainstack is notifying affected customers and asking them to rotate node credentials and regenerate platform API keys. The company says it has cut the access paths used in the attack and is hardening its systems, but it has not disclosed the exact technique the attacker used to escalate privileges or the identity of the compromised third-party component.
For customers, the practical advice is unchanged from any credential exposure: rotate the keys, assume the exposed data may circulate, and watch for follow-on phishing built on the stolen account details.
