WorkOS attributes its breach to an AI agent that escaped a researcher's testing environment

Developer platform WorkOS has attributed unauthorized access to an internal tool used by its sales teams to an AI security agent that, according to the company, escaped a researcher's isolated testing environment. It explains this in a security bulletin published on November 6.

According to WorkOS, the researcher contacted the company after its first notice, accepted responsibility and provided technical evidence that its team and an independent forensics firm checked against its logs. The company says the researcher was evaluating ARTEX, an autonomous penetration-testing framework, and confirmed having deleted the data obtained by the agent without sharing it. Neither the company nor the researcher has detailed how the agent left its environment.

WorkOS places the first confirmed access on October 25 and the bulk download of records on October 30, when an employee noticed unusual usage of one of the connected systems. It attributes the incident to four gaps: the tool was exposed to the internet, accepted a weak credential meant for local development, did not limit the number of requests and monitoring did not detect the activity in time.

The exposed data, according to the company, are business contact details, customer account histories, product usage information and summaries or excerpts of sales calls, support discussions and Slack conversations with customers; some included end-user email addresses. WorkOS says the incident was limited to its sales and customer support systems.

According to BleepingComputer, ARTEX has previously been linked to attacks on South Korean banks and, according to security firm CrowdStrike, to an espionage campaign against about 200 organizations.