Is Shadow AI Controllable?

I’ve been noticing at work regardless of any tools that’s being used to block ChatGPT or Claude etc, my coworkers are naturally finding ways around it, even resorting to taking picture so they can ask ChatGPT on their phone.

Nothing malicious at all, in their defence they’re just trying to be productive because internal AI tools are rubbish, but it did make me wonder if the real problem isn’t “how we block block users from pasting into chatgpt”, but rather, “how do you enable secure use of AI without hampering productivity.”