What happened
Netskope announced (Sept 15, 2026) Skylight Agent Action Control, a new capability that classifies every AI agent action into nine intent-based risk categories (e.g. data destruction, exfiltration, RCE, credential manipulation) and blocks high-risk actions before execution, running on existing inspected network traffic with no new console/agent required. It ships as part of a renamed 'Netskope Skylight' AI security portfolio (formerly Netskope One AI Security) and will be generally available end of the current quarter.
Why it matters
Provides deterministic, pre-execution policy enforcement against 'authority drift' in agentic AI — addressing IDC-cited data that 91% of orgs cannot currently stop a risky agent action before it executes and 54% reported an AI agent security incident in the past year.
Applicability
Enterprise security/SASE teams already on Netskope One should evaluate ahead of end-of-quarter GA for agent governance policy; broadly relevant to any org deploying coding/RPA/browser agents.