Varonis reported the flaw to Google in late 2025 and it has been addressed, but it reminds defenders to take a fresh look at their AI Infrastructure security.

This is an ingested summary. Full reporting is available at the primary source below.