What happened
Replaying the AWS cryptomining campaign, November 2025
- 1
Stolen IAM credentials
From Nov 2, 2025The credentials of an IAM user with admin-like privileges are used from an external hosting provider, and the actor probes EC2 quotas and permissions; how the credentials were obtained is not stated.
- 2
Roles and workloads
From Nov 2, 2025The actor creates a service-linked role and a Lambda role, dozens of ECS clusters that run a Docker Hub miner image, and EC2 launch templates and Auto Scaling groups that start mining at boot.
- 3
Mining and persistence
From Nov 2, 2025Miners run within 10 minutes of access and the actor sets termination protection on every launched instance; The Register also reports a public Lambda Function URL kept for access, which the part of the AWS post that was read does not describe.
Within 10 minutes of access to an account the actor had miners running on ECS Fargate and on EC2 Auto Scaling groups, with termination protection set on the launched instances; AWS describes a coordinated campaign across several customer accounts, ongoing at publication; the costs to victims, the number of accounts and the amount mined are not stated.
Where cloud-provider controls fit
On AWS, a service control policy that accepts calls only from the organization's networks would have refused calls from an external hosting provider. A service control policy that refuses IAM user calls without MFA would have bounded the stolen credentials to sessions that carry MFA; the sources do not say whether the credentials were an access key or a console password. With access, service control policies that keep the creation of roles, containers and compute, and attribute changes on instances, to approved administrators would have limited who could start the miners and set termination protection; AWS describes the compromised user as having admin-like privileges, and no source says whether it would have been among those administrators. A permissions boundary on new roles would have capped what those roles could do. An EC2 declarative policy that sets VPC Block Public Access would have left the miners no path to the pools, though the compute would still be billed until removed. The public registry image is outside the provider's reach, and how the credentials were obtained is not stated.
Could this have been prevented?
In hindsight, not as a whole. 2 controls would have refused a step outright, and 8 controls would have limited one. 1 step has no cloud-provider control. The counts below are control assessments, not attack stages: several can address the same step. The same replay with the controls in place before the incident shows where each phase would have been refused or bounded.
- 2
- Prevented
- 8
- Contained
- 1
- Not covered
How it could have been prevented
The same replay with the controls in place. Green marks where a step is refused, blue where it is bounded, and red where no control reaches it. Open a row to read the control, why it acts, and its caveat.
Replaying the AWS cryptomining campaign, November 2025
1 · Stolen IAM credentials
1 Prevented2 Contained4 on other clouds2 · Roles and workloads
4 Contained1 Not covered9 on other clouds3 · Mining and persistence
1 Prevented2 Contained6 on other cloudsKey facts
- AWS dates the campaign start to November 2, 2025 and described it as ongoing in its report of December 16; GuardDuty engineers found similar techniques across several customer accounts, and The Register says a handful of accounts were alerted
- The actor used compromised IAM user credentials with admin-like privileges from an external hosting provider; no AWS service flaw was exploited, and how the credentials were obtained is not stated
- After calling GetServiceQuota and RunInstances with DryRun set to test permissions, the actor created a service-linked role for Auto Scaling and a Lambda role, then dozens of ECS clusters, sometimes more than 50, with tasks of 16,384 CPU units and 32,768 MiB that ran the Docker Hub image yenik65958/secret:user
- The actor created 2 launch templates whose UserData started mining at boot and 14 Auto Scaling groups with a maximum of 999 instances and a desired capacity of 20, and called RunInstances directly after the Auto Scaling quotas ran out
- Miners were operational within 10 minutes of initial access, per AWS as quoted by The Register; the image had more than 100,000 pulls, and the miner ran SBRMiner-MULTI against pools on port 17155
- The actor set disableApiTermination on all launched instances, so responders must re-enable termination before deleting them; BleepingComputer reports that customers were alerted and told to rotate credentials and that the Docker Hub image was removed
Sources
Every fact in this breakdown comes from the public reporting below. The replay is a model of that reporting, not a reconstruction of the environment it describes.
- 1.Cryptomining campaign targeting Amazon EC2 and Amazon ECSAWS Security Blog · Dec 16, 2025 · Vendor reportTimeline, credentials, attacker steps, API calls, ECS and EC2 scale, termination protection, indicators, detection
- 2.Crypto crooks co-opt stolen AWS creds to mine coinsThe Register · Dec 18, 2025 · NewsSummary of AWS findings, Lambda function URL persistence as reported, AWS recommendations as reported, what is not stated
- 3.Amazon: Ongoing cryptomining campaign uses hacked AWS accountsBleepingComputer · Dec 17, 2025 · NewsTimeline as reported, image and scale as reported, customer notification
The estate in the replay is illustrative and drawn at the level of attack stages; the victim's real environment is not public. Verdicts describe what a control does when it is in place before the incident. They are not an assessment of the victim's security program.