How boards can respond to emerging AI risks

The AI risks boards should be addressing now

WTW's latest Directors’ and Officers’ Survey found that 56% of respondents consider AI a very or extremely important risk. Yet the challenge extends beyond the technology itself. Different organisations are concerned about different aspects of AI, from misinformation and fraud to the risk of falling behind competitors. Only 55% of respondents also believe their boards have the skills and knowledge needed to provide effective AI oversight.

Understanding these concerns can help you focus governance efforts where they matter most.

AI-generated misinformation

Half of respondents identified AI-generated errors and misinformation as one of their primary AI concerns.

Generative AI can produce content that appears authoritative and credible while containing inaccuracies, missing context or incorrect conclusions. While AI hallucinations are well known, the challenge is becoming more nuanced.

As large language models (LLMs) continue to improve, obvious mistakes are becoming less common. The remaining errors may be harder to identify because they sit within otherwise high-quality outputs. A report, analysis or recommendation can appear entirely reasonable while containing subtle flaws that influence decision-making.

To reduce this risk, you should establish clear validation processes for AI-generated information. Human review remains essential, particularly where outputs inform strategic decisions, customer communications or regulatory reporting. Additional AI tools may also support quality assurance activities, but human oversight and accountability must be intentionally designed and integrated into such activities to avoid complacency and algorithmic bias.

AI-enabled fraud

Forty percent of respondents cited AI-enabled fraud and social engineering as a significant concern.

The growing availability of synthetic media has made it easier to create convincing digital content. Images, videos, audio recordings and written communications can now be generated quickly, cheaply and at increasingly high levels of quality.

For threat actors, this creates opportunities to strengthen traditional fraud techniques. AI can support phishing campaigns, impersonation attempts and highly targeted social engineering attacks at a scale that was previously difficult to achieve.

Technology controls remain important. Multi-factor authentication, monitoring tools and strong cyber security practices continue to provide important protection. However, improving AI literacy across your organisation is equally valuable.

When employees understand how AI-generated content can be created and manipulated, they are often better equipped to recognise suspicious communications and challenge unusual requests.

Strategic adoption risks

Not all AI risks stem from using the technology. Thirty-eight percent of respondents expressed concern about failing to adopt AI effectively.

This is ultimately a strategic issue. Organisations that overlook relevant AI opportunities may miss efficiency gains, new capabilities and competitive advantages that others are already beginning to realise. The challenge is that there is no universal AI strategy. The most relevant opportunities depend on your sector, operating model, growth ambitions and risk appetite.

You should begin by identifying where AI could have the greatest impact on your organisation. Questions worth exploring include:

• How could competitors use AI to strengthen their position?
• Which business activities could change most significantly?
• What skills will your workforce need in the coming years?
• Where could AI deliver measurable value?

Answering these questions can help prioritise investment and focus attention on areas where AI presents the greatest opportunity or risk.

Strengthening board oversight

One finding stands out from the survey. Only 55% of respondents believe their boards possess the skills and knowledge needed to oversee AI effectively.

AI now touches strategy, operations, cyber security, governance, workforce and reputation. Boards don’t need to become technical specialists, but they need enough understanding to challenge assumptions, oversee risk and make informed decisions.

For many organisations, board-level AI education is a practical starting point. Building familiarity with AI capabilities, limitations and governance considerations can improve oversight and support better decision-making.

Some organisations may also benefit from bringing additional AI expertise into board or advisory structures, particularly where AI plays a significant role in business strategy or operations.

Building confidence in AI governance

The AI risks highlighted in our survey are diverse, but they share a common theme: the need for informed oversight and effective governance. From misinformation and fraud to competitive pressures and board capability, organisations face difficult decisions about how they adopt and manage AI.

Those that invest in governance, controls and AI literacy today will be better positioned to manage emerging risks while capturing the opportunities AI presents in the years ahead.

To discuss how your organisation can strengthen its approach to AI risk, governance and resilience, contact WTW's Risk & Resilience Advisory team.