AI risk is already hitting the P&L. What do CFOs do next?

With AI risk impacting business performance, making governance and accountability a financial imperative rather than a technology issue should be top of mind for finance leaders.

In brief

• AI risk is already producing measurable financial losses, with 98% of surveyed organisations reporting AI-related losses and many experiencing impacts exceeding US$1 million. 
• Traditional finance controls are often ill-equipped for AI because AI systems are adaptive, less predictable and sometimes opaque, making risks harder to identify, manage and value.
• Finance teams are uniquely positioned to help make AI trustworthy at scale by combining enterprise-wide visibility, governance discipline and oversight structures that support both value creation and risk management. 

AI risk is showing up on the balance sheet, but just 11% of chief financial officers can match key AI risks to the appropriate controls. Every unmanaged risk carries a financial signature.

Most finance leaders are thinking hard about how AI might help them move faster, work more efficiently and unlock better insight. There’s no shortage of interest or information – but there is a shortage of time.  

What’s harder – particularly for time-poor CFOs – is creating the space to step back and think about the implications. 

The latest EY Responsible AI Pulse Survey shows what’s at stake: 98% of surveyed companies report financial losses linked to AI-related risks. Two-thirds have experienced losses of more than US$1 million, with an estimated average loss of US$4.9 million. 

Here’s how these losses show up. Time and capital sunk into pilots that don’t scale. Rework thanks to inaccurate or poor-quality training data. New cybersecurity vulnerabilities from poorly governed systems. Compliance and remediation costs linked to regulatory breaches. Legal exposure arising from biased AI-driven decisions.  

CFOs are accustomed to managing risk in environments where accuracy, integrity and control matter. But AI doesn’t behave like the systems finance teams know how to govern. Traditional controls assume stable inputs, predictable behaviour and transparent logic. AI systems are adaptive and, in some cases, opaque. That makes risk hard to see and harder to price. 

This is happening alongside an already crowded agenda. Compliance, regulation, sustainability, transformation and capital allocation continue to demand attention, alongside the reality of manual, month-end processes that still must get done. Against that backdrop, finance leaders are being asked to build digital literacy in a fast-moving and unfamiliar domain.

But the pressure to upskill is real. AI is being embedded in forecasting, reporting and decision-support processes. Used well, it has the potential to free finance teams from backward-looking work and shift effort towards forward-looking decision support. 

Finance teams aren’t always seen as enablers. With this technology, we can be. Finance sits across an organisation, isn’t siloed, has access to enterprise data and has the discipline to translate information into insight that supports better decisions. If any team is well positioned to make AI trustworthy at scale, it’s finance. 

The EY survey data points to where value is being realised. Organisations that have put the right governance in place are more likely to see improvements in the areas CFOs care most about, like revenue growth and cost savings. Those with formal oversight structures report stronger performance across both areas.  

For CFOs, the next move isn’t to slow AI down or speed it up – it’s to make it governable. By recognising that every AI-influenced decision already carries a financial signature, if finance can’t explain how an AI-driven decision was made, can they be comfortable signing off on the result?