Up to 78 per cent of organisations have already implemented AI models into their core operations, according to F5‘s latest State of Application Strategy (SOAS) report. Significantly, IT leaders are increasingly opting for full control of the technology at the expense of apparent convenience. Only 8 per cent of enterprises rely exclusively on off-the-shelf, public AI-as-a-Service. The rest prefer to build and develop their own diversified environments.
However, this shift in strategy comes with a huge operational challenge. Today’s business operates in an extremely distributed model, with as many as 93 per cent of organisations operating in multi-cloud environments. In such environments, managing an average of seven different artificial intelligence models simultaneously becomes a major focus for technical directors. Kunal Anand, head of product at F5, notes that the key question is no longer ‘if’ but ‘how’ to deploy AI reliably and securely at scale. More than three-quarters of companies are now focused on generating real answers and results, demonstrating the market maturity of projects.
– AI has already passed the experimentation stage and has become part of the standard operation of companies. Today, the question is no longer whether companies will use AI, but whether they will be able to do so reliably, securely and at scale, points out Kunal Anand, Chief Product Officer at F5.
However, the shift from theory to day-to-day production has exposed serious vulnerabilities. As many as 88 per cent of organisations admit that they have already experienced AI security incidents. This challenge will become even more urgent as business stands at the threshold of the next wave of agentic AI. Almost all of the companies surveyed (98 per cent) are preparing for the implementation of these autonomous systems, which require a completely new approach – their own controls, permissions and digital identity – to operate smoothly.
As a result, the cyber security landscape is undergoing a fundamental transformation. Traditional protection at the physical infrastructure level is giving way to management within the applications themselves. A new critical point for engineers is becoming the control of prompts, tokens and APIs. Nearly a third of organisations point to query (prompt) management as the most important element of AI delivery.
