It's been almost four years since OpenAI launched ChatGPT, bringing generative artificial intelligence (AI) into the mainstream. In that time, AI has moved from a novelty to a valuable business tool. According to McKinsey's 2025 State of AI survey, 88% of organisations were using AI in at least one business function in 2025. This is up 10% from 2024. The same survey found that 23% of organisations are scaling an agentic AI system somewhere in their business and an additional 39% have started experimenting with AI agents. On top of this, Gartner forecasts that worldwide spending will reach $2.7 trillion in 2026, a 49.5% year-over-year increase.
While this growth is exciting and there’s no doubt that AI opens up massive possibilities, adoption is accelerating faster than the policies around it. This gap is where risk lives, says Jonathan Oaker, founder and CEO of CloudZA. Without proper AI governance, businesses risk compromising sensitive customer/commercial data, which could lead to regulatory breaches, fines and reputational damage. But when you’re trying to put the right processes and frameworks in place, it’s easy to see governance as a handbrake.
“The goal of governance is not to hamper innovation or slow businesses down, but rather to make sure that organisations can access the insights offered by AI without sharing any intellectual property or personal information,” he says. “We're seeing organisations rush to deploy AI without understanding what data it's accessing, who's accountable for its outputs or whether it complies with regulations.” If your business uses well-known frontier tools like ChatGPT and Claude without governance, your data is probably being used to train these models and make them smarter.
Imagine if a major retailer with a strategic approach to stock management feeds sales and inventory data into ChatGPT's free plan. If the model uses this information for training, competitors could access confidential information about the brand’s forecasting or replenishment strategies and use this data to erode the retailer’s competitive advantage. “Even if you have a paid subscription, it’s important to check your data settings and disable the option that allows your conversations to be used to improve the model. Paid plans don’t all handle data in the same way. Good governance means understanding the terms of the service for the model you’re using.”
From a governance perspective, visibility supports accountability and makes it easier to enforce internal policies. Without sound governance, it’s hard to monitor AI use, spot unusual activity and track operational costs. With the necessary governance in place, businesses can access a snapshot of all AI interactions. “This is stored in the business and allows you to go back to a conversation and see what data was or wasn’t shared. It’s possible to build this visibility into more open models, but it isn’t easy.”
Governance isn’t a cheap journey, and it certainly isn’t the easier route, Oaker concludes. “If you look only at upfront costs, it’s far cheaper to ignore it than it is to prioritise it. But if something goes wrong and you don’t have the necessary frameworks in place, the results can be disastrous.”
Understanding the risks of poor AI governance is only the first step. The next is knowing how to put the right safeguards and processes in place. In the follow-up to this press release, CloudZA will explore the practical moves businesses can make to get their AI governance on track.
For more insights into AI governance, watch this podcast recorded live at the AWS Summit in Johannesburg.
Share
