Skip to content

Urgent Escalation Request: Account Compromise & $2000+ Fraudulent Bedrock Charges

1

Hello AWS Community and Support Team,

I am requesting urgent assistance to escalate an active support case regarding a severe account compromise on child account.

Summary of the Incident:

  • Incident: Account compromise via stale domain takeover resulting in $2,000+ unauthorized Amazon Bedrock charges.
  • Immediate Containment Taken: Applied a global 'Deny *' Service Control Policy (SCP) to completely isolate the child account, verified access denial, and secured the root email.
  • Detection: Detected immediately via AWS Budget alerts ($100 threshold).

Request: Due to the high fraudulent charge amount and imminent billing cycle, I am kindly asking an AWS Moderator to help flag and escalate my active Support Case ID ************ to the AWS Trust & Safety / Billing team for an immediate payment hold and billing waiver investigation.

Thank you!

*Edit: Removed case ID — Aimee K.

2 Answers
0

Hi there,

I understand how unsettling it must be to discover that your account has been compromised, this is definitely a stressful situation.

I've raised your concerns right away to our Support team so we can investigate and take appropriate action.

Please continue monitoring your support case, that's where you'll receive all important updates and next steps.

— Aimee K.

AWS
EXPERT

answered 9 days ago

EXPERT

reviewed 9 days ago

0

The subdomain takeover vector here is documented in AWS's Security Blog (Threat tactic spotlight article). It happens when a DNS CNAME points to a deleted resource in a global namespace like S3, CloudFront, or Elastic Beanstalk. An attacker claims that resource name in their own account, your DNS resolves to their resource, and they gain access through that vector.

The $100 budget alert caught this after the damage was done. Two things would have detected it much earlier. First, GuardDuty has an AI Protection finding specifically for anomalous Bedrock token volumes that might indicate a cost harvesting attack (grounding score 0.98 against the official GuardDuty docs). It learns the baseline token volume per identity and alerts when invocations deviate significantly. Second, AWS Cost Anomaly Detection uses ML to flag spending pattern changes regardless of absolute dollar amount, so it would have alerted on the sudden Bedrock usage where none existed before.

The preventative fix is simple: delete DNS records first, wait for TTL expiry, then delete the resource. There's also an open source AWS Config custom rule (aws-samples/sample-dangling-dns-detection) that scans Route 53 for dangling CNAMEs and generates Security Hub findings automatically.

Your SCP isolation response was correct for containment. If you want a full security pillar assessment to identify these gaps systematically and get a prioritized remediation roadmap, I do AWS Well-Architected Reviews focused on exactly this kind of post-incident hardening. . Otherwise the AWS Security Blog article and that GitHub sample are solid starting points to prevent recurrence.

I do AWS security and architecture advisory work.

answered 9 days ago

EXPERT

reviewed 9 days ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.