← All posts

NewsDLPDSPMInsider Risk Managementdata protection

Island Extends Enterprise Browser Controls to AI Agents

Island, the enterprise browser vendor that raised a $400M Series F at a $6.4B valuation earlier this month, announced a product expansion that brings AI agents into its governed browser sessions alongside human users. The new capability lets organizations connect autonomous agents to SaaS applications, internal tools, and public websites while applying the same governance stack Island already uses for people: identity federation, data boundaries, credential isolation, and full session attribution.

Island CTO Dan Amiga framed the move as a natural extension of controls “proven in production” for human browsing, now applied to machine actors that read, click, and act on behalf of users. That framing matters for the DLP and DSPM crowd watching the space: as copilots and agentic workflows get write-access to internal apps and sensitive data stores, security teams need a way to see and restrict what those agents can access and exfiltrate — without waiting for agent frameworks themselves to ship native data-loss controls.

The announcement reinforces a thesis Island and peers like LayerX and Menlo Security have been pushing for a couple of years: the browser is the highest-leverage control point for modern data protection, because it sits between the user (or now, the agent) and virtually every SaaS and web-based data source. Rather than retrofitting DLP policies into each agent orchestration platform, Island is betting that enforcing data boundaries at the browser layer generalizes cleanly whether the actor typing into a web app is a person or an LLM-driven agent.

This is a product-direction announcement, not new funding — Island’s Series F was already widely covered — but it’s a useful signal of where enterprise browser vendors are taking DLP-adjacent controls next. Expect competitive responses from other secure-browser and SSE vendors as agentic AI access to SaaS and internal systems becomes a mainstream insider-risk and data-exfiltration concern heading into 2027 budget cycles.