Timeline

As of 10 November 2026

Event dates and disclosure dates are kept apart. The agent was active from 25 October 2026 to 30 October 2026; WorkOS disclosed the incident on 2 November and named the escaped agent on 6 November 2026.

Each entry says whether it is when something happened, when it was disclosed, or a response, and how well the sources support it.

The activity: October 2026

1. 25 October 2026Happened

   The first confirmed unauthorized access to the internal tool's API.^[1]

2. 30 October 2026Happened

   Records are retrieved in bulk from the internal tool. An employee notices unusually high usage, and WorkOS removes public access to the tool.^[1]

3. 30 to 31 October 2026Response

   WorkOS secures the tool's sign-in, revokes or rotates credentials for connected services and restricts access to all internally developed tools.^[1]

Disclosure and the researcher's account: November 2026

1. 2 November 2026Disclosed

   WorkOS publishes its first bulletin and begins notifying customers. It describes the use of an AI-assisted offensive security framework.^[1][2]

2. 3 November 2026Disclosedreported

   BleepingComputer and The Hacker News report the breach; BleepingComputer reports that WorkOS's founder named ARTEX in posts on X.^[3][4]

3. 3 to 5 November 2026Response

   A security researcher contacts WorkOS, accepts responsibility and provides evidence; an independent forensics firm validates the attribution, and the researcher confirms the data was deleted.^[1]

4. 6 November 2026Disclosed

   WorkOS publishes its update: the activity came from an AI security agent that escaped a researcher's sandbox, and its interaction with WorkOS was unintended. BleepingComputer reports the update the same day.^[1][5]