In our managed SOC we have been tracking a pattern: actors who compromise an environment early, then go quiet. They are not exfiltrating for show — they are collecting live sessions, tickets and cached tokens to reuse weeks later, when the initial footprint has been cleaned up and their presence no longer triggers alerts.
Why this one is hard to spot
- The credentials are real — a replay looks identical to legitimate use
- The activity is slow and spread thin, sitting below typical volume thresholds
- Detection depends on identity context, not just “is this a known bad IP”
Because the login itself is valid, most perimeter tooling simply does not flag it. That pushes the defence deeper into identity monitoring — session anomaly detection, unusual source geography for a given account, and correlation across systems rather than point-in-time checks.