What happened
Replaying the Codefinger January 2025 intrusion
- 1
Standing credentials: keys from the internet
In the weeks before Jan 13, 2025 (dates not stated)The actor uses AWS access keys that were publicly disclosed or otherwise compromised and that allow s3:GetObject and s3:PutObject. No AWS vulnerability is exploited. How the keys were obtained is not stated.
- 2
Impact: objects re-encrypted with a key only the actor holds
In the weeks before Jan 13, 2025 (dates not stated)The actor encrypts objects with SSE-C, using an AES-256 key it generates and stores locally. AWS does not store the key, and CloudTrail logs only an HMAC of it. AWS describes large numbers of CopyObject calls with SSE-C that overwrite existing objects.
- 3
Impact: a seven-day deletion clock and a ransom note
In the weeks before Jan 13, 2025 (dates not stated)The actor sets lifecycle rules through the S3 Object Lifecycle Management API to mark files for deletion within seven days, and leaves a ransom note in each affected directory with a Bitcoin address and a client ID. Whether the keys held a lifecycle permission is not stated, and neither is whether any victim paid or any data was deleted.
Objects in the S3 buckets of two organizations were re-encrypted with a key only the actor holds, and lifecycle rules were set to delete them within seven days; Halcyon states that there is no known method to recover without the key. The sources do not say what the data was, whether any victim paid or whether any data was deleted.
Where cloud-provider controls fit
Every step of this intrusion uses valid keys and ordinary S3 calls, and AWS cannot tell valid use of valid credentials from malicious use. The controls act before and around those calls. A service control policy that refuses new long-term access keys limits the keys that can be disclosed to those that already exist, and a service control policy with a network condition refuses a key used from the internet. A bucket that blocks SSE-C rejects writes that specify SSE-C while the block stands, but a key that holds s3:PutEncryptionConfiguration lifts the block, and the sources do not say whether these keys held it. A resource control policy that denies the SSE-C header bounds the same write across the organization, although AWS's resource control policy pages do not name that condition key. If the re-encryption still happens, S3 Object Lock in compliance mode keeps an unencrypted copy for you to restore, and a service control policy keeps non-administrators from writing the lifecycle rule. The ransom note is an allowed write: no cloud-provider control refuses it, and whether you hold a copy to restore stays your decision.
Could this have been prevented?
In hindsight, no control would have stopped it. 7 controls would have limited how far it went. 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.
- 0
- Prevented
- 7
- 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 Codefinger January 2025 intrusion
1 · Standing credentials: keys from the internet
2 Contained3 on other clouds2 · Impact: objects re-encrypted with a key only the actor holds
3 Contained3 on other clouds3 · Impact: a seven-day deletion clock and a ransom note
2 Contained1 Not covered6 on other cloudsKey facts
- Halcyon reported two victims in its January 13, 2025 research; SecurityWeek attributes the figure of two to an AWS spokesperson, and the dates of the activity are not stated
- The actor used AWS keys that could run s3:GetObject and s3:PutObject; no AWS vulnerability was exploited, and how the keys were obtained is not stated
- The objects were encrypted with SSE-C under an AES-256 key that the actor generated and stored locally; AWS does not store the key and CloudTrail logs only an HMAC of it, so Halcyon says there is no known method to recover without it
- AWS describes large numbers of CopyObject operations with SSE-C that overwrote existing objects, while Halcyon names the s3:PutObject permission; the sources differ on the operation
- The actor set lifecycle rules to mark the files for deletion within seven days and left a ransom note with a Bitcoin address and a client ID in each affected directory; whether any victim paid or any data was deleted is not stated
- On April 6, 2026 AWS announced that S3 disables SSE-C by default for new general purpose buckets and for existing buckets in accounts with no SSE-C encrypted objects
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.Abusing AWS native services: ransomware encrypting S3 buckets with SSE-CHalcyon (RISE Team) · Jan 13, 2025 · ResearchAttack sequence, SSE-C encryption, lifecycle deletion, ransom note, victim count, recommendations, unknowns
- 2.Preventing unintended encryption of Amazon S3 objectsAmazon Web Services (AWS Security Blog) · Jan 15, 2025 · Vendor reportAttacker operations, overwrite by CopyObject, recommended controls, bucket policy and resource control policy example, valid credentials used in an unintended way
- 3.Ransomware abuses Amazon AWS feature to encrypt S3 bucketsBleepingComputer · Jan 13, 2025 · NewsAWS statement, attack summary, recovery impossibility
- 4.Compromised AWS keys abused in Codefinger ransomware attacksSecurityWeek · Jan 14, 2025 · NewsVictim count origin, AWS statement, attack summary
- 5.Amazon S3 starts rolling out new security best practice to new and existing buckets by defaultAmazon Web Services (What's New) · Apr 6, 2026 · Vendor reportLater provider change, SSE-C disabled by default
- 6.Blocking or unblocking SSE-C for a general purpose bucketAmazon Web Services (S3 User Guide) · undated · Vendor reportBlocked encryption types setting, request behavior, existing objects, later provider change
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.