Training before access – a new control layer for AI?
Jinfo Blog

By Anja Chemnitz Thygesen

Abstract

Item

Training end users in AI as a control layer

Instead of treating training as a one-off awareness exercise, organisations are increasingly using it as a practical control layer.

In some environments, end users are required to complete AI training before gaining access to approved tools or advanced functionality.

This reflects a broader shift in how organisations think about enablement.

Rather than simply rolling out AI tools and hoping policies will be followed, organisations are beginning to introduce more structured models:

• baseline AI literacy
• tool-specific training
• progressive access to more advanced capabilities.

The logic is straightforward. If organisations expect end users to handle licensed content responsibly, they must first ensure users understand:

• what AI tools can and cannot do.
• what rights apply to content.
• where the risks emerge during everyday work.

This is particularly important because many end users still assume that access to licensed content automatically includes the right to use that content in AI systems. In practice, these are often separate rights.

Connecting learning systems with access management

Some organisations are therefore connecting learning systems directly with access management. Completion of training can trigger access to approved AI tools, while users who have not completed training may remain restricted to basic functionality.

This approach has clear advantages in several ways. It:

• creates a visible governance framework.
• helps organisations demonstrate due diligence.
• reinforces the message that AI access comes with responsibilities – not just convenience.

At the same time, organisations are discovering that this quickly becomes operationally complex.

Managing who has access to which tools – and under what conditions – can be cumbersome, particularly as the number of approved AI solutions grows.

One information manager described the challenge bluntly:

"It's very hard to keep up with who has access, what they're using, and how they're using it."

A delicate balance between control and usability

Restricting access is also a balance, as too tight restrictions may make users simply move towards external tools outside organisational visibility.

Providing broad access without sufficient guidance, on the other hand, means a significant increase in risk of misuse.

The organisations making the most progress appear to focus less on "policing" and more on "reinforcement". Training is treated as an ongoing process, supported through follow-ups, practical examples, targeted interventions, and workflow-based guidance.

The best approach is no longer simply writing AI policies.

It is creating environments where responsible behaviour becomes part of everyday work – and where access, guidance, and governance are increasingly connected.

Read the full analysis and practical recommendations:

• "Training end users in AI – from policy to practice" (Jinfo Report)

« Blog