Netskope Launches AI Agent Action Control to Preempt Risky Automation — Netskope, Inc.

Event summary

Netskope introduced Skylight Agent Action Control on September 15, 2026, classifying AI agent actions into nine risk-based categories before execution.
The feature allows security teams to block, allow, or alert on actions based on predefined risk profiles for different agent types.
91% of organizations currently lack the ability to stop risky agent actions before execution, per Netskope's 2026 AI Risk and Readiness Report.
Netskope rebranded its AI security portfolio as 'Skylight,' consolidating five existing products under this umbrella.

The big picture

As enterprises accelerate AI agent adoption, probabilistic security controls are proving insufficient. Netskope's deterministic approach positions it as a potential standard-bearer for governing autonomous systems. The rebranding underscores the company's pivot toward comprehensive AI security, a $12B+ market projected to grow at 35% CAGR through 2030.

What we're watching

Adoption Pace: How quickly enterprises will integrate Agent Action Control into existing security frameworks, given 54% reported AI agent incidents in the past year.
Competitive Response: Whether rival security vendors will accelerate development of similar preemptive AI governance tools.
Regulatory Alignment: The extent to which Netskope's deterministic controls will influence emerging AI safety standards.