GitBook says an attack “driven largely by automation and AI” used a flaw in its single sign-on to reach its production systems

GitBook, the documentation platform, has published a post-mortem on an attack on its systems. It says an attacker used a flaw in its single sign-on login flow to create sessions acting as GitBook staff, then queried a misconfigured GitHub App integration that reached internal code repositories and a Cloudflare API key. That key led to production secrets and full infrastructure access through a service account.

GitBook says “the attacker moved with a speed and precision that pointed to an attack driven largely by automation and AI.” The attacker had access to its production environment for about 2:45 hours on the evening of 2 November and about 40 minutes early on 3 November, the post-mortem says.

GitBook disabled the service accounts and rotated all secrets, including customer secrets it manages. It says billing information, passwords and IP addresses were not compromised, and that the attacker's activity focused on a single customer, but it treats customer data such as tokens and integration secrets as exposed. The page does not name the attacker.

The Switchboard ran WorkOS's account of a breach by an AI agent on 7 November. This story is in the catch-up lane: the post-mortem covers 2–3 November and had not run.

• Claimed GitBook says the attack was driven largely by automation and AI. GitBook
• Confirmed The attacker had production access for about 2:45 hours on 2 November and about 40 minutes on 3 November, GitBook's post-mortem says. GitBook
• Claimed Billing information, passwords and IP addresses were not compromised, and the activity focused on a single customer. GitBook