Businesses are losing millions due to unmanaged AI risks at every level, according to new research from EY, with employees creating their own ‘shadow’ AI agents and C-suites lacking the knowledge to implement appropriate controls. 

Polling almost 1,000 executives around the world, including 100 from the UK, EY’s latest Responsible AI (RAI) Pulse survey found that almost every company (99%) reported financial losses from AI-related risks, with 64% experiencing losses over $1 million (£752k).

Bang on trend, 98% of UK firms said that unmanaged AI risks had cost them financially, with more than half (55%) losing over $1 million, and the average losses standing at a whopping $3.9 million (£2.9m) per firm.

EY found the most common AI risks suffered by UK firms include non-compliance with regulations (57%), inaccurate or poor-quality training data (53%) and high energy usage impacting sustainability goals (52%), figures in line with global averages – showing that there remain systemic flaws across enterprise AI adoption.

The survey found that time may be running out to find solutions, as AI risks are expected to skyrocket with the broad deployment of agentic AI.

These advanced AI systems, which can autonomously make decisions and take purposeful actions without human input, are becoming more prevalent in the workplace, but EY found caution in their deployment.

Most UK organisations (81%) said they continuously monitor their agentic AI processes and models to ensure they adhere to responsible AI principles, while 80% said they have incident escalation procedures in place in case an AI agent behaves unexpectedly.

But those guardrails may not be enough to protect against the unknown AI agents lurking in the network. EY’s report found a rising trend in ‘citizen development’, employees independently creating or deploying AI agents.

While a third (32%) of companies have chosen to prohibit the practice outright, two-thirds of UK firms (64%) allow this activity, with 53% implementing formal policies to ensure alignment with responsible AI principles.

However, companies admit they lack visibility into employees’ AI activity, creating what EY describe as a governance gap where shadow AI development can flourish undetected, meaning they are essentially flying blind.

The problem is compounded by a lack of knowledge among the C-suite, with many executives not knowing how to apply the right controls to mitigate AI risks.

When asked to match the appropriate controls against five AI-related risks, only 12% of global respondents got them all right, though UK leaders did slightly better at 17%.

Unsurprisingly, CIOs and CTOs performed the best, though still only around a quarter answered correctly across all five use cases, while Chief Risk Officers (those ultimately responsible for AI risks) lagged at 11% on average around the world. 

The cost of this gap in awareness is steep, with companies that suffered losses exceeding $10 million (£7.5m) typically having fewer than half of the recommended safeguards in place, averaging just 4.5 out of 10. 

Recommended reading

• 88% of UK Execs Would Trust AI To Handle a Crisis
• UK Orgs Set to Increase AI Investment by A Third by 2026
• AI Saves Developers 10+ Hours A Week – But Inefficiencies Cost Them The Same

In contrast, firms with losses under $1 million had implemented 6.4 controls on average, a disparity underscoring the need to equip leadership with the skills and knowledge to manage AI effectively, especially as the financial and reputational stakes continue to climb.

“UK companies that see responsible AI as a strategic advantage instead of an overhead will lead the pack,” said Matthew Ringelheim, EY UK&I AI and data leader.

“They will build trust both within and beyond their organisation and accelerate speed to market — bringing the latest technologies into production ahead of their competitors. 

“As organisations continue to navigate the complexities of AI integration, prioritising responsible governance will be essential for driving sustainable growth and maintaining a competitive edge in the market.”