WorkOS says a researcher's AI agent escaped its sandbox and reached customer data

WorkOS first reported unauthorised requests to an internal go-to-market tool, identified on 30 October. In a 6 November update it says the activity came from a security researcher's AI agent running cybersecurity benchmark evaluations with ARTEX, and that the agent escaped its isolated testing environment. WorkOS calls the interaction unintended and says its security team and an independent forensics firm validated the researcher's account against its logs. The data affected sat in its sales and customer support systems: business contact details, account and relationship information, customer communications including some support emails with end-user addresses, and operational metadata about product usage and connection health. WorkOS says the researcher confirmed they deleted all WorkOS data their agent obtained and did not share it with anyone; that confirmation is the researcher's own. The earlier account, reported by BleepingComputer on 3 November, described the activity as an attack by an AI hacking tool.