UltraViolet Cyber

9 out of 10 organizations we studied have no way to detect automated attacks against their AI systems. That's one finding from 640 structured observations across 10 enterprise security programs. The full benchmark covers 64 specific activities. See how 10 real programs scored across each one.

UltraViolet interviewed security, AI and engineering leaders across 10 organizations and scored their programs against a shared catalog of 64 activities. Every organization has approved AI coding assistants. Only one has built detection for automated attacks against AI systems. Use the findings to gauge where your own program sits.

Encryptionless extortion now accounts for a quarter of tracked attacks. Backups don't solve it. When attackers steal your data and threaten to leak it, your recovery posture is irrelevant. The leverage is the data itself. @Dan Gittis, Director of UltraViolet Threat Intelligence, on the attack chain, the groups running it, and what it changes for your defenses.

Can you name every AI system influencing decisions in your environment? 12 questions reveal where your governance program stands—including shadow AI.

One Ollama configuration change can expose an unauthenticated API to the public internet. UltraViolet’s TIDE Team explains the risk.

67% of employees access AI services through non-corporate accounts on work devices. UltraViolet’s John Waller explains why traditional security controls can’t see it.