This rule detects a sequence in AWS CloudTrail where a single long-term IAM user access key (AKIA*-prefixed) enumerates Bedrock foundation models via ListFoundationModels and then immediately invokes a model within the same 15-minute window. The pattern—model discovery followed by invocation using the same long-term key—is unusual in production, where Bedrock workloads typically run under IAM roles with short-lived credentials. This combination is consistent with LLMjacking: an adversary using stolen credentials to discover AI capabilities and run high-cost model inferences at the account owner's expense. The rule correlates two events (ListFoundationModels and InvokeModel or related actions) on the same access key within 15 minutes, using fields like aws.cloudtrail.user_identity.access_key_id, aws.cloudtrail.user_identity.arn, aws.cloudtrail.request_parameters, and source metadata to validate the sequence and the invoked model. It maps to ATT&CK techniques for Cloud Service Discovery (T1526) and Valid Accounts (T1078.Cloud Accounts) under Initial Access, highlighting the risk of credential abuse to access AI services. The rule is designed to trigger on legitimate Bedrock usage only if it follows normal patterns, and highlights the importance of migrating to IAM roles and restricting long-term keys for Bedrock workloads.