The board’s role in AI governance

The board’s role in AI governance: Where do directors fit in this crucial new corporate responsibility as it continues to evolve?

Artificial intelligence will be one of the few things that define the 2020s. For businesses, it’s perhaps the most important development because of its massive potential to alter the way we work.

It goes without saying that something as game-changing as AI carries with it equally strong levels of risk. Because of that, proper governance of AI is essential. However, the proper implementation of AI governance principles is, as of 2026, alarmingly limited. 

While 99% of chief executive officers report active or planned AI deployments, up to 79% of corporate directors admit to having minimal or no technical knowledge around them, according to Deloitte’s 2024 board oversight analysis. 

Without a doubt, this is a serious capability deficit. To practically address it, AI has to form part of the governance strategy, with rules, overseers, and metrics, just as with any other area of governance. 

AI governance is important because a company’s ability to manage AI is rapidly becoming central to its ability to remain healthy. AI constitutes opportunity, risk, and a major boardroom topic. Companies cannot afford to leave it off the agenda. 

2026 figures from tech-management solutions firm Ivanti say that, among IT professionals, 27% identify governance, security or compliance concerns as their organisation’s biggest AI deployment obstacle. That’s above issues like skills shortages (20%) and tech limitations (17%). In other words, the people who work on the frontlines with AI are now calling for a rulebook and strategy more than they’re calling for anything else. 

Unstructured, ill-thought-out, and poorly-led AI projects suffer from extraordinary failure rates. Research from 2026 indicates that more than 80% of enterprise AI initiatives fail to deliver their promised business value, while 42% of global companies abandoned the majority of their AI projects in 2025 due to spiralling costs and issues with data quality. Moreover, an estimated 95% of generative AI pilots fail to generate any net positive return on companies’ profit and loss statements. 

See the trend? While the willingness to adopt AI might not be a problem for a lot of companies, governance capabilities certainly are, because the results simply don’t indicate that companies are following the same governance standards that we’d expect in other areas of business.

When it comes to analysing the board’s role in detail, we need to look at two things: where boards should engage, and where they should step back. There must be a clear boundary between the two in order to preserve the board’s top-down view. 

Of course, executive directors’ work will straddle this boundary, but for everyone else, it will be more clear-cut.

• Setting strategic goals and limits: Directors should discuss and sign off on the goals they want to see the company reach with AI. As part of this, they should also set a threshold for risk levels (and understand the different kinds of AI risk in the process, like algorithmic error, intellectual property exposure and data privacy).  
• Oversee a strong data foundation: Directors should establish how they will measure the effectiveness and risk around AI, implementing independent audits where necessary to address data quality, cybersecurity concerns, etc. 
• Ensure consistency: This should be done, at least in the first instance, through the C-suite. Directors should use this communication channel to monitor how AI strategy is being implemented. If necessary, they can approve new roles in this area. For example, they could introduce one or both of a C-Suite-level chief AI officer reporting to an AI board committee.

• Model management and design: Directors shouldn’t be as involved as making daily or weekly decisions about specific algorithmic designs, neural networks, or vector database choices. These belong strictly to IT/tech teams. However, you must ensure that important communications make it from those teams to management and the board. 
• Data operations: Day-to-day tasks related to data management, building pipelines, and cleaning datasets are firmly in “salaried employee” territory and not something for the board to look after. 
• Procurement: Companies may well end up using multiple AI models, but selecting vendors, negotiating contracts, and adjusting workflows based on the company’s needs should all be left to management. However, the board should have some involvement in the final sign-off of major contracts, and the responsibility to ask big-picture questions about those contracts if things don’t make sense.  

Boards will need to keep their high standards of oversight as AI becomes more central to how a company functions. To do that, they need to think about who sits at the table, and whether they have the skills to provide meaningful leadership from an AI context. 

Data from Deloitte shows that, in 2025, 40% of companies were thinking differently about their boards’ make-up because of AI. It’s a good sign in some ways: AI is at least commanding a significant slice of attention, but realistically, that figure should be higher, given AI’s potential impact.  

The central point of the evolution will be Agentic AI – systems that can execute multi-step workflows with minimal human input. Many tools of this kind are still in the testing phase, but as the years go on, there will be greater availability in the open market, and new avenues for your company to explore. 

Directors don’t need to learn the tech specifics about every agentic model, but they need to constantly ask the big questions about them: what are the risks? Where is the human oversight? How are our products affected?

The board’s role in AI governance is a careful balance between the following core principles:

• Boards must engage with the risks and opportunities of AI. 
• Boards must actively oversee the implementation of solid AI governance structures and strategy. 
• Boards must not allow themselves to get too bogged down in day-to-day matters (except for certain executive directors); same as with other areas of governance. 
• Boards have a responsibility to ensure they devote enough time, resources and have enough skills to a thorough AI governance framework that works both short and long term.

• Governance of AI: A Critical Imperative for Today’s Boards
• AI Project Failure Statistics 2026: The Complete Picture
• AI Project Failure Rate 2026: 80% Fail
• Corporate Governance in the Context of Business Digital Transformation
• AI Governance Consulting: How to Choose a Partner (2026)
• 2026 AI Business Predictions
• Gartner Predicts by 2027, 50% of Enterprises Without a People‑Centric AI Strategy Will Lose Their Top AI Talent
• Gartner D&A Summit 2026: Key Takeaways on Context & AI