Press play to start listening
AI is a hot topic for every business. You can’t avoid it, whether CEOs are discussing how to use it to improve productivity or the general public is turning to AI to get quick answers to everyday questions.
For businesses, the appeal is obvious. AI-powered tools can save time, automate repetitive tasks, and make everything from meetings to content creation more efficient – a quick scroll through Instagram and you’ll see a plethora of restaurants using it to advertise their menus, for example.
But as more businesses introduce AI into their everyday workflows, another question is becoming increasingly important: can people actually use these tools safely?
In the news, there have been a lot of reports on AI safety recently. As you can imagine, there are going to be AI tool adoption barriers, which is why we look into why executives won’t adopt what they can’t control when it comes to AI.
Visibility is the first challenge
One of the biggest concerns for executives is knowing exactly how AI is being used across an organisation.
Employees may already be using consumer AI platforms to summarise documents, write emails, analyse data or troubleshoot technical problems (we’ve all heard of ChatGPT). If these tools are being used without the knowledge of IT or security teams, organisations can end up with what is effectively an unmanaged layer of technology.
This creates questions around data: What information is being entered into an AI system? Where is that information stored? Who can access it? And what happens to it afterwards?
Without visibility, security teams cannot properly assess or manage the risks.
Control needs to be built in
For AI adoption to scale, businesses need more than policies telling employees what they can and cannot do. They need technical controls that make safe behaviour easier. This could include restricting which AI tools employees can access, controlling what data can be shared, monitoring AI activity, and establishing clear permissions for AI agents interacting with internal systems.
The same principle applies to AI-powered automation. An AI agent that can send emails, access customer records or update business systems needs clearly defined boundaries. Executives are far more likely to approve these technologies when they can see what the AI is allowed to do – and, importantly, what it is prevented from doing.
Security cannot be an afterthought
Traditional software is generally designed around predictable processes. AI systems can behave differently because their outputs can vary depending on the information and instructions they receive.
That makes governance particularly important. Security teams need to be involved before AI tools are deployed rather than being brought in after a problem occurs. They should help establish acceptable use policies, access controls, auditing processes, and mechanisms for stopping or restricting AI activity when necessary.
This doesn’t mean preventing employees from using AI. In fact, excessive restrictions can encourage people to find their own workarounds. Instead, the goal should be controlled adoption.
Overall, the future of workplace AI won’t simply belong to the companies using the most advanced models. It will belong to those that can make AI powerful and controllable. For executives, that distinction could be the difference between approving an AI pilot and rolling it out across the entire organisation.
Which AI tools are you currently using and why? Will you be trying a new one shortly?
