This is GuardDuty’s correlation engine. Instead of looking at a single event in isolation, it connects multiple signals across services and time.
What it is designed to find
- Multi-stage attacks that span data sources, resource types, and time.
- Correlated behaviors rather than single isolated anomalies.
- Attack sequences that may involve IAM, EC2, EKS, ECS, or S3-related compromise patterns.
Memory hook: Standard GuardDuty finds suspicious events. Extended Threat Detection finds the story those events tell together.
Weak signals
A weak signal is an activity that may look harmless by itself, but becomes suspicious when combined with other actions.
- One unusual API call may not trigger a major finding.
- Several related actions across time can reveal reconnaissance, privilege escalation, or data compromise.
Exam shortcut: Weak signal alone may not matter. Weak signals + findings + time correlation = attack sequence.
24-hour rolling time window
Extended Threat Detection uses a rolling time window to identify sequences of activity that suggest an in-progress or recent attack. This matters because sophisticated attacks do not always happen in one moment.
1
Reconnaissance: suspicious listing or discovery actions.
2
Access or privilege change: credentials or permissions shift in a risky way.
3
Impact: resource misuse, data access, or workload abuse appears.
Scope reminder
Extended Threat Detection correlates signals within a single AWS account. It does not correlate activity across multiple AWS accounts or across an AWS Organization as one combined attack graph.
Exam shortcut: Organizations help centralize management and visibility, but Extended Threat Detection still evaluates each account independently.
Suppression rule impact
Archived findings created by suppression rules are not considered in attack sequence correlation.
That means noisy suppression rules can hide signals that might have helped identify a bigger attack pattern.