Attackers can publish public AWS machine images with names that closely resemble trusted images, hoping that developers or automated deployment pipelines choose them without checking the publisher. If selected, a malicious image can introduce malware, exposed credentials, or insecure settings into a Golden AMI and spread the risk across many cloud workloads.
How the Attack Worked
The whoAMI attack relies on a simple trick: an attacker publishes a public AWS Machine Image whose name closely resembles a trusted vendor image. When a developer or an automated CI/CD pipeline searches for an AMI by name alone, without validating the owner ID, it can select and deploy the malicious image instead of the legitimate one. Because public AMIs can be published by anyone, the name or description alone offers no guarantee of trustworthiness.
Why It Succeeded
This technique succeeds because deployment workflows often prioritize convenience over verification. Searching by image name is fast, but it skips the step of confirming who actually published the image. Once a compromised AMI is selected, it can be promoted into a Golden AMI and replicated across accounts, regions, and production workloads, meaning the same risk is inherited by every instance derived from it. Public and misconfigured AMIs may also carry malware, vulnerable software, hardcoded credentials, SSH keys, or insecure settings, compounding the impact once the image is in wide use.
What to Watch For
- AMI names that closely resemble a familiar or trusted vendor image
- Images sourced from unknown or unapproved publishers
- Deployment workflows or pipelines that search for AMIs by name without checking the owner ID
- Golden AMIs that have not been rebuilt or reassessed recently
- Unexpected AMI creation, modification, sharing, or launch activity in logs
How to Build Resistance
Organizations can reduce exposure to look-alike image attacks with a few concrete controls:
- Always define and verify the AMI owner ID before approving or deploying an image; never rely on name or description alone
- Use AWS Allowed AMIs to enforce deployment from trusted, pre-approved image sources
- Treat Golden AMIs as supply-chain dependencies: assess them before promotion and rebuild them on a regular schedule, since even well-maintained images become outdated otherwise
- Monitor AWS CloudTrail for AMI creation, modification, sharing, and launch events
- Detect and investigate EC2 instances launched from unknown or unapproved AMIs
Key Takeaway
The whoAMI pattern shows how a routine action, selecting an image by its familiar name, can become a supply-chain entry point when ownership verification is skipped. Building owner-ID checks and approved-image policies directly into pipelines removes the reliance on trust in a name, closing the gap that this attack technique depends on.
Key findings
- The whoAMI attack involves publishing malicious public AMIs whose names closely resemble legitimate vendor images.
- Developers and CI/CD pipelines can select a malicious AMI when they search by name but do not validate the image owner.
- A compromised source AMI can be turned into a Golden AMI and replicated across accounts, regions, and production workloads.
- Public and misconfigured AMIs may contain malware, vulnerable software, hardcoded credentials, SSH keys, or insecure settings.
- Checking publisher identity and AMI owner IDs is a key control against look-alike image attacks.
Who’s being targeted
- Commonly targeted roles: Developers, DevOps engineers, Cloud engineers, Platform engineering teams, CI/CD pipeline owners, Cloud security teams, IT change-management teams.
- Affected industries: Organizations using AWS EC2, Technology and software development, Cloud-based enterprises.
- Attack channels: website.
- Impersonated: Trusted AMI vendor or publisher.
Red flags to watch for
- The AMI name resembles a trusted image but the owner ID is not verified.
- The image comes from an unknown or unapproved publisher.
- The deployment workflow searches by image name alone.
Frequently asked questions
What is the whoAMI attack?
It is an attack where adversaries publish malicious public AWS AMIs whose names closely resemble legitimate vendor images, hoping developers or CI/CD pipelines select them based on name alone.
How does a malicious AMI spread across an organization?
If a compromised source AMI is turned into a Golden AMI, it can be replicated across accounts, regions, and production workloads, spreading malware, exposed credentials, or insecure settings.
How can teams defend against look-alike AMI attacks?
Teams should verify the AMI owner ID and publisher identity rather than trusting an image name, use AWS Allowed AMIs to enforce trusted sources, and monitor CloudTrail for unexpected AMI creation, sharing, or launch events.
Who is most at risk from this attack technique?
Developers, DevOps and cloud engineers, platform engineering teams, CI/CD pipeline owners, and cloud security teams at organizations using AWS EC2 are the primary targets.
Read the video transcript
That familiar AMI name? It could be a copy. In the whoAMI attack, someone publishes a public AWS image with a trusted vendor’s name. Search by name alone, and you or your CI/CD pipeline could select it. Here’s the trap: a familiar name isn’t a verified publisher. If that image becomes your Golden AMI, hidden malware or credentials can spread across accounts and production. Before approving any AMI, check its Owner ID against your approved publisher list.