Critical Docker Hub Flaw Lets Attackers Impersonate Users and Push Container Images

Docker has disclosed a critical Docker Hub vulnerability that could let attackers impersonate other users and publish container images under trusted names. Tracked as CVE-2026-105445, the flaw affected personal access tokens transferred during account conversions.

Docker resolved the issue on September 30, 2026, stating its investigation found no evidence of exploitation. This issue posed a software supply chain risk, as attackers could potentially alter images that developers relied on.

Depending on the token’s permissions and the target user’s access, they could pull, push, or delete images in personal repositories and connected organizations. However, the flaw did not allow full account takeover.

Docker Hub Vulnerability

Docker Hub allows users to convert a personal account into an organization while keeping existing repository names. This helps teams preserve image references used by applications and build systems.

During conversion, the user selects an existing Docker Hub account to become the organization’s owner. Starting August 26, 2024, this process also transferred the original account’s personal access tokens, known as PATs, to the selected owner. Docker changed the ownership record without changing or revoking the token secret. The new owner was neither asked to approve the transfer nor notified that it happened.

An attacker could therefore create an account, generate a PAT, and convert that account into an organization while naming a victim as its owner. The attacker still knew the token secret, but Docker Hub now linked that token to the victim. Registry requests using it could authenticate as the target user.

This was a failure in credential-ownership handling rather than a stolen-password attack. The token remained usable after its identity changed. Docker says the exposure window lasted from August 26, 2024, at 20:33 UTC until September 30, 2026, at approximately 19:10 UTC, when the production fix went live.

Publishing an image under a legitimate repository name could expose downstream users to unexpected code.

Cybersecurity News previously covered the Trivy supply chain attack involving compromised Docker images, illustrating why trusted publishing paths matter. That incident is separate from this vulnerability, which Docker found no evidence attackers exploited.

The affected tokens did not provide access to the Docker Hub website. Attackers could not use this flaw to change a victim’s password, email address, or two-factor authentication settings.

The impact stayed at the registry level and depended on the token’s scope and the user’s repository permissions. Researchers Peter Molnar and Krisztian Hozak of mptrdev.com reported the issue through Docker’s vulnerability disclosure program.

According to Docker’s security advisory, the company reproduced it with internal test accounts and deployed a fix less than four hours after receiving the report.

Docker reviewed every account conversion since the flaw appeared and checked registry authentication logs, usage data, and token records. Investigators examined approximately 304,000 image pushes made with transferred tokens.

All went to repositories belonging to the converted organization or the new owner, with none reaching unrelated accounts. Docker also inspected higher-risk public images without running them. It found no malware, cryptominers, credential theft, persistence, or tampering.

However, registry logs record token use rather than token existence. Docker therefore treated every transferred PAT as potentially affected, including tokens that never appeared in usage logs.

Docker notified exposed users and organization owners on October 2. On October 8, it deactivated transferred tokens, except those owners requested to retain. Account conversion no longer transfers PATs.

Docker is also adding owner consent, notifications, and records of token ownership changes. Docker says users who received no notification were not exposed and need no incident-specific action.

It recommends revoking unfamiliar or unused PATs and using organization access tokens for organizational automation. Related container image scanning tools provide context for teams reviewing image security alongside credential controls.

An EDR alert reveals an attack. But can it tell the full story? Discover how to reconstruct the attack timeline.

The post Critical Docker Hub Flaw Lets Attackers Impersonate Users and Push Container Images appeared first on Cyber Security News.