Google has fixed a Cloud Platform (GCP) security flaw that could allow hackers to backdoor user accounts through malicious OAuth applications. Astrix Security, the cybersecurity firm that discovered the vulnerability, named it GhostToken. Attackers could exploit the flaw by linking a malicious app to a user’s OAuth token, which gave access to their Google account.
They could then make the app invisible to the user, making it impossible for them to revoke access.
Attackers could hide the app from Google’s application management page, which is the only place Google users can manage apps connected to their accounts.
To hide the malicious app, hackers could put it in a “pending deletion” state by deleting the linked GCP project, and then restoring the project to obtain a refresh token to retrieve a new access token. The process could be repeated in a loop to gain access to the user’s data repeatedly.
The attack’s impact depended on the level of permissions granted to the malicious app installed by the victim. The vulnerability allowed hackers to gain permanent and unremovable access to a user’s Google account by converting an already authorized third-party application into a malicious trojan app.
Google’s patch addresses GCP OAuth applications in ‘pending deletion’ states, and they will appear on the ‘Apps with access to your account’ page, allowing users to remove them and protect their accounts from hijack attempts.
Astrix Security advised all Google users to visit their account’s app management page and check all authorized third-party apps, ensuring that each app only has the permissions it needs to function. The GhostToken vulnerability was reported to Google by Astrix Security in June 2022, and the patch was released globally in April 2023.
