Observed incident

Symantec traces AI-assisted credential theft and lateral movement at a U.S. retailer

Reported

Published here

What happened

In September, Symantec observed an intruder at an unnamed U.S. retailer switch from failing manual commands to scripts that reflected reasoning about earlier results. The investigation covered the retailer’s web server and at least six additional servers.

Parties and reporting

Where this happened

Affected:
United States

Countries and regions are approximate and come from the sources, not from attacker, vendor or researcher locations. See this report’s footprint on the world map.

AI tools and systems: AI-generated scripts (model unidentified), Windows servers

Outcome

The attackers obtained credentials and moved to another host. Investigators observed no theft of business data or encryption in this intrusion.

Operating lesson

Review linked activity across web, management and database servers when blocked commands give way to successful credential access.

Evidence and limits

Symantec assessed AI assistance with high confidence and human direction with moderate confidence. It could not identify a model. A possible link to a September 22 report on another retail operation remains unconfirmed.

Sources