The EU AI Act Blind Spot: Browser-Level Governance

European compliance teams are mapping AI systems, classifying risk levels, and building documentation trails the EU AI Act demands. This is serious work. Almost all of it targets AI tools that organisations formally procure and deploy.

The far harder problem, said Matt Smith, Field CTO, EMEA at Island, is governing how employees actually use AI in real time. Most employees access generative tools, coding assistants, and autonomous agentic AI through a browser tab. Sensitive business data moves into these tools routinely. Often there is no awareness of where that data is retained or who can access it.

Regulation is starting to catch up. The EU AI Act is the most significant attempt yet to place formal obligations around transparency, auditability, and data handling. Some headline deadlines have slipped, reflecting the difficulty of governing AI in practice. The compliance questions it raises -- who uses which AI tools, what data they handle, how to demonstrate oversight -- are ones every board should be asking, Smith said. Most organisations cannot answer them.

Compliance efforts cover approved and procured AI systems. They rarely account for the shadow AI problem: unsanctioned tools employees adopt independently, accessed through a browser session that leaves no audit trail and sits outside most security architectures. Agentic AI tools -- those that take actions autonomously rather than responding to prompts -- make this more urgent. They move data at speed and scale that makes after-the-fact detection redundant.

To understand the governance gap, Smith said, look at how the working day has changed. For most employees, the browser is no longer just a tool for accessing work. It is where work happens. SaaS platforms, cloud collaboration tools, financial systems, HR applications, and AI tools all live inside browser tabs.

Security architecture has not made the same journey. The controls most organisations depend on were built for a world of corporate networks, managed devices, and applications sitting behind a firewall. That model does not reflect reality. It still underpins how most businesses approach security. The result is a blind spot: most controls establish whether a user is permitted to reach an application, Smith said. They have no view of what happens once the user is inside it.

That gap matters beyond compliance. When an employee copies a customer record into a personal email, pastes financial data into an external AI platform, or downloads a sensitive report to an unmanaged device, none of those actions triggers an alert. The access was authorised. The behaviour was not governed.

Organisations have responded by adding tools -- data loss prevention solutions, cloud access security brokers, virtual desktop infrastructure, VPN layers. Each addresses a specific gap. Together they form a stack that is expensive to run, difficult to manage, and still leaves the browser session itself largely ungoverned, Smith said. The problem was not a shortage of tools. It was building on the wrong foundation.