JINX-2401 is a threat actor that has been observed by Wiz Threat Research while targeting AWS Bedrock with compromised credentials and attempting to hijack LLM models in multiple AWS environments.
Their techniques included creating new IAM users with policies granting Bedrock permissions and setting up console profiles for completing the LLM agreement process. The attacker relied on API calls such as PutUseCaseForModelAccess and CreateFoundationModelAgreement to establish access but was blocked by SCPs. Despite gaining Administrator Access permissions in one environment, multiple attempts to invoke models failed. The campaign exhibited consistent behavioral patterns, including naming conventions for IAM users and policies, which facilitated cross-environment detection.