Vulnerabilities · Application Security
Docker Hub flaw could have let attackers impersonate users and tamper with images
A critical vulnerability in Docker Hub allowed attackers to impersonate users and alter container images. Docker says it found no evidence of abuse before fixing the flaw.
Ross Grant3 min read
The short version
A bug in Docker Hub meant attackers could pretend to be other users and change or delete their container images. Docker fixed the problem quickly and says it did not find any signs that attackers used this flaw. Only users who got a warning from Docker need to do anything.
By the numbers
- 304,000pushes with transferred tokens reviewed by DockerBleepingComputer
- 2024-08-26date vulnerability was introducedBleepingComputer
- 2026-09-30date Docker received the vulnerability reportBleepingComputer
- <4 hourstime taken by Docker to fix the flaw after confirmationBleepingComputer
What happened?
A critical vulnerability in Docker Hub (CVE-2026-105445) allowed attackers to impersonate users and access their container repositories using personal access tokens. Docker fixed the flaw in September 2026 and says it found no evidence of abuse. Affected users were notified and compromised tokens deactivated.
What the vulnerability allowed
A critical flaw in Docker Hub (CVE-2026-105445) made it possible for attackers to impersonate other users and access their container repositories. This included the ability to pull, push, or delete images, depending on the permissions of the compromised account. The issue was linked to how Docker Hub handled personal access tokens (PATs) when converting a personal account into an organisation.
How the attack could work
According to BleepingComputer (opens in a new tab), an attacker could create a Docker Hub account and a PAT, then convert the account into an organisation and assign a target user as the owner. Due to the flaw, Docker Hub would reassign the attacker’s PAT to the new owner (the target) without changing the secret or asking for the target’s consent. This meant the attacker could use their original PAT to act as the target user at the registry level.
Depending on what the PAT allowed and the permissions of the target, the attacker could pull, push, or delete images in any repositories the target could access. This included potentially publishing malicious images under a legitimate user’s name, which Docker described as a software supply chain risk.
Timeline and response
The vulnerability was introduced on 26 August 2024, when Docker Hub began reassigning PATs during account conversion without revoking or rotating the secret. The issue was reported to Docker on 30 September 2026. Docker confirmed the problem and deployed a fix within four hours. The company says it found no evidence of exploitation after reviewing 304,000 pushes made with transferred tokens and analysing high-risk public images.
Docker notified affected users and organisation owners on 2 October 2026. On 8 October, all transferred PATs were deactivated unless owners specifically requested to keep them. Docker now requires user consent before making someone an organisation owner and no longer transfers PATs during conversion.
What was not affected
Docker clarified that the flaw did not give attackers access to Docker Hub web accounts, nor did it allow them to change passwords, email addresses, or two-factor authentication settings. This limited the risk of account lockout or broader compromise.
What users should do
Docker says users who did not receive a notification are not affected and do not need to take action. As a general practice, users should review their access tokens and permissions, and be cautious about accepting ownership roles in organisations.
What happens next
The flaw has been fixed and Docker has changed its processes to prevent similar issues. However, the incident highlights the importance of careful token management and user consent in cloud platforms. Security teams should monitor for unusual activity in their repositories and review any automated workflows that use PATs.
Claimed vs confirmed
- Confirmed
Attackers could impersonate users and tamper with images via Docker Hub flaw.
Docker confirmed the vulnerability and its potential impact, according to BleepingComputer.
- Confirmed
No evidence of exploitation was found.
Docker says it found no signs of abuse after reviewing logs and images.
The impact
- Money
- No financial losses or fines have been reported. However, under GDPR, a breach could result in fines up to €20 million or 4% of annual turnover if personal data were involved.
- Reputation
- Docker's reputation could be affected due to the critical nature of the flaw, but the quick response and lack of evidence of abuse may limit long-term damage.
- Customers
- Users whose accounts were affected were notified and had their tokens deactivated. Others do not need to take action. There is no evidence that any images were tampered with or that users need to change passwords.
Why it matters
Docker Hub is widely used to share and run software containers. If attackers can tamper with trusted images, they could spread malicious code to many users and organisations. This is a serious supply chain risk for anyone who relies on Docker images.
How does it work?
The flaw allowed attackers to create an account and personal access token (PAT), convert the account into an organisation, and assign a target user as owner. Docker Hub reassigned the PAT to the new owner without changing the secret or asking for consent, letting the attacker use it to act as the target user.
Key points
- A Docker Hub flaw allowed attackers to impersonate users using personal access tokens.
- Attackers could have tampered with or deleted container images in affected accounts.
- Docker fixed the issue quickly and found no evidence of abuse.
- Only users notified by Docker were affected; others do not need to act.
- The incident highlights the importance of token management and user consent.
Sources
BleepingComputer
Reference
