AI concerns: ROI, scaling and governance – part 2
In part 1 of our AI concerns: ROI, scaling and governance article series, we heard from the ITIL community about their current issues with, and approaches to handling, the AI governance challenge.
In part 2 of the series, we explore the impacts of where AI governance has gone wrong, the gap between AI adoption and workforce readiness and - again - we invite further viewpoints from the ITIL community, including how they are using guidance from the ITIL AI Governance (Version 5) certification to help them manage risk, scale adoption responsibly and connect AI investment to business value.
The recent cyber-breaches by AI agents created by high-profile companies including OpenAI and Anthropic were, at the time, described as “rogue”.
However, the term has since been dismissed by some commentators who claim the work of the AI agents was anything but rogue. Quoted in the Financial Times (FT), Boyan Milanov, senior research scientist at the independent, New York-based AI Now Institute, said: “The offensive capabilities we have reached today, we have reached deliberately.”
Artificial intelligence Editor at the FT, Madhumita Murgia, called it a “turning point in global cyber security” as “AI agents are now able to string together different and complex skills to attack real-world targets without outside control or help.” This, she adds, makes “improved safeguards all the more important”.
But what is the current state of employee capability when it comes to using, managing and governing AI?
Research in the UK by QA - and reported in the HR Director publication - suggests there is already a gap between adoption of AI and who has the “confidence and capability to use it well”.
QA’s chief learning officer, Jo Bishenden, said: “Organisations need to treat capability as something to track deliberately, the same way they’d track any other rollout, rather than assuming it improves on its own once the technology is in place. The gap between adoption and capability doesn’t close by itself – it closes because someone is paying attention to it.”
AI: speed, operational improvements and increased profitability
Despite the warnings about AI threats and skills gaps, practitioners within the ITIL community are witnessing value and ROI generated by the technology:
Walter Macuada, a regional consulting leader, said: “Some of our services now incorporate agentic capabilities across secure software development, quality assurance and deployment, as well as vulnerability analysis and remediation. This has enabled us to accelerate delivery, improve operational efficiency, reduce manual effort, and increase the profitability of our services.”
ITIL consultant, trainer and Ambassador, Patricia Edith Fridman, noted how AI is helping improve productivity, accelerate research and content development, and design more tailored consulting and training materials. “The value is already visible in reduced preparation time and faster response, although we are still developing more formal metrics to quantify the return on investment,” she said.
The term “AI-native” - a central pillar of the ITIL (Version 5) guidance and certification scheme - is one used by Mike Rossiter, senior consultant at technology consultancy, Enablis, to describe his company’s current direction of travel, meaning it “can build fast and also niche products for our clients [including] charities, large financial services and gaming organisations”, while adding that “DevOps practices apply more than ever before to cope with this speed of adoption”.
ITSM manager, Wajahat B, highlights how AI and predictive analytics in service management can help reduce incidents, improve service uptime, and enhance overall service quality. “We are also working on implementing AIOps and AI-based agents to further improve service management through better automation, faster issue detection, and more proactive service operations,” he added.
To scale or not to scale AI: accountability, human judgement and meeting governance demands
However, setting aside the initial benefits of AI experienced by ITIL practitioners, taking the technology across an enterprise remains a distinct challenge.
“The main difficulties are protecting confidential client information, ensuring the quality and accuracy of AI-generated outputs, developing consistent AI skills and selecting use cases that produce measurable value,” Fridman explained.
Integrating AI into existing ways of working without weakening professional accountability, critical thinking or human judgement was another issue, she said, along with “the ability of professionals to continually update their knowledge and skills so that we can offer clients the most relevant, reliable and effective solutions”.
The challenge of realizing value from AI at scale has led, according to Wajahat B, to his organization introducing “more stringent AI screening and assessment, to ensure proposed solutions are technically feasible, integrate with our environment, meet governance and regulatory requirements, and can deliver the capabilities and outcomes promised before significant investment is made”.
AI governance approaches: from “work in progress” to a Governance Board and screening process
Constructing new approaches to manage and govern AI in their organizations is clearly keeping ITIL practitioners busy.
Feisal Ismail, a principal consultant, described how his company is formulating guardrails and AI policies surrounding the impact of AI use on data privacy, security and ethics in the context of the business, though this - he said - remained a work in progress.
“Every solution”, consulting director, Sunil Kumar, explained “will go through a proper development cycle, console-based monitoring to track the progress, regular validations of the solution and capability-aligned governance” which is already sustaining the solutions and reducing the cost of service for customers.
Wajahat B’s organization has gone further, establishing a “fully-fledged AI Governance Board”, along with an AI screening process at the initial stage before any financial commitment is made.
He said: “The screening helps ensure that requirements are clearly understood, the proposed solution is properly assessed, and the investment is estimated as accurately as possible. This early assessment also helps identify any additional infrastructure, integration, licensing, security, or regulatory requirements upfront, reducing the risk of unexpected costs later and providing a clearer view of the total cost of ownership.”
Organizations adopting a holistic approach to AI governance is something also cited by Walter Macuada, who said: “We are establishing an AI governance framework that combines policies and responsible-use guidelines with defined roles, risk assessments, human oversight, and security and privacy controls. We are also working on maintaining visibility of AI use cases and agentic solutions, evaluating them throughout their lifecycle, and monitoring both the value delivered and the associated risks.
“Training and awareness are also key measures, particularly to reduce uncontrolled or shadow AI adoption and promote the safe and responsible use of AI across the organization.”
ITIL AI Governance: harnessing the benefits of technology for business
As revealed in part 1 of this article series, more than 50% of ITIL community members had already adopted guidance from the new ITIL AI Governance certification, with more than 80% consulting 2025’s ITIL AI Governance white paper.
And several practitioners shared how the guidance has already contributed to their AI governance approaches:
Aleksandar Dimitrov pointed to the guidance on strategic alignment and readiness assessment: “It helped us avoid ‘AI for the sake of AI’ by ensuring every implementation directly drives tangible business goals rather than just technical outputs.”
How ITIL AI Governance bridges technology and business imperatives was reiterated by Patricia Fridman, who noted: “The emphasis on connecting AI initiatives with organizational objectives, defining clear accountability and applying proportionate governance according to risk is particularly useful. It reinforces the importance of balancing innovation with human oversight, ethical use, transparency and continual evaluation.”
And the guidance’s “practical approach to starting with clear, high-value, initial use cases instead of attempting large-scale AI adoption immediately,” was welcomed by Allan Soares of the Regional Labour Court of the 10th Region in Brazil, adding that “it also helped us recognize that traditional IT governance does not fully address AI-specific risks”.
Finally, the ITIL AI Capability Model - sometimes referred to as the 6C model - was dubbed “the most useful guidance” by Walter Macuada: “We have used it to assess the scope and implications of AI use across our services. This has helped us identify where AI can deliver value, understand its impact on services and stakeholders, and recognize the risks and governance considerations associated with each use case”.
Human oversight is the vital element in managing and governing the risks, purpose and desired results from AI initiatives. Therefore, investing in training and knowledge to close the professional and organizational skills gap may be one of the most important decisions a business makes this year.
Build practical AI governance capabilities with ITIL
Learn how ITIL AI Governance can help you assess AI capabilities and risks, integrate governance into existing ways of working and scale AI responsibly while maintaining human oversight and alignment with business objectives.