
AI
Malware Targeting AI Infrastructure: Threats and Defense
Adversaries compromise model weights and training data to alter behavior, creating threats that traditional signature-based security tools cannot detect or block.
Everything Sandbox Desk has reported about malware that targets AI infrastructure: 2 stories, newest first. Part of our AI coverage.

Adversaries compromise model weights and training data to alter behavior, creating threats that traditional signature-based security tools cannot detect or block.

Standard endpoint protection misses attacks that steal model weights or corrupt training data through legitimate API calls.