So the attacker turns logging off (was log file validation enabled? usually isn't in Terraform ) which does not fire an alert because there is no alerting. Then does their bad stuff ... Then modifies the logs (which are in an S3 bucket on the compromised account, remember!) Then they turn logging on? The whole point is alerts go outside AWS. They go to like, your inbox or pagerduty or whatever. If they had no alerts then what use are their logs, which could have been modified? Do you think they set up cross-account logging or had enable_log_file_validation set to true?