US firms say AI is outpacing governance, survey finds

ABBYY has published research showing that 60% of US businesses say AI is developing faster than their organisations can govern it. The survey also found that only 22% of US companies completely trust AI systems to operate without introducing unacceptable risks.

The findings are based on a study of 1,200 senior managers at companies with more than 100 employees across the US, UK, France, Germany, Australia and Singapore. The research highlights concerns over accountability, limited confidence in AI outputs, and continued friction between investment and returns.

US respondents reported higher spending and greater confidence in internal governance than peers in other countries. Even so, they described a gap between the pace of AI adoption and the systems meant to oversee it. The study found that 90% of US organisations said they had adequate systems for auditing AI for compliance, fairness and transparency, while 91% said data governance and ownership clarity were adequate.

That confidence did not extend to trust in the technology itself. Just over a quarter of business leaders globally said they completely trust AI systems to produce accurate outputs. The same share said they completely trust AI to explain how important decisions were reached or to protect confidential information. Confidence was lower on risk, with only 22% saying they completely trust AI systems to operate without introducing unacceptable risks.

Responsibility split

The study found no clear consensus on who should be accountable when AI systems produce harmful or incorrect outputs. Nearly a third of leaders at businesses using AI said responsibility should be shared between the organisation and the vendor.

Another 26% said the organisation using the AI should bear primary responsibility, while 20% said it lies fully with the AI vendor. A further 12% placed responsibility on the end user, and 10% said regulators should take the lead.

The lack of alignment also appears inside organisations. Senior leadership was seen as having the clearest understanding of who is responsible for developing, implementing and managing AI, with respondents saying 89% of senior leaders know where responsibility sits. That fell to 84% for middle management and 65% for junior employees, although the US figure for junior staff was higher at 79%.

Data and returns

Data governance emerged as a major constraint on AI deployment. Seven in 10 business leaders surveyed said data governance requirements slow AI deployment, suggesting concerns over accuracy, security, and compliance still limit how quickly companies can move projects into wider use.

The study linked those constraints to financial outcomes. Less than half of global business leaders at organisations using AI said their AI initiatives had exceeded expectations, though the figure was higher in the US at 53%.

Data quality was cited as the biggest barrier to achieving greater returns from AI, with 20% naming it as their top challenge. Integration with existing systems followed at 15%, with implementation costs at 13%.

US companies continue to spend more. The survey found that US organisations spent an average of USD $925,000 on AI tools in the past 12 months, up from USD $794,000 the previous year, an increase of 16%. Respondents said they expect investment to rise by a further 21% next year, compared with a global average of 19%.

Governance frameworks

Many organisations said they already have formal structures in place to manage AI. In the US, 85% of businesses said they have an AI governance framework, and 83% said it has made AI initiatives more successful.

The picture was less complete for specific incident controls. Half of global organisations said they have an AI incident reporting process, while 46% have an AI incident response plan. Only 35% have an AI kill switch, and the same proportion have AI rollback arrangements, although both figures rise to 41% in the US.

Human review remains part of decision-making in many companies. Globally, 56% of business leaders said humans review all important AI decisions before action is taken, rising to 60% in the US.

Ethical concerns are also affecting deployment speed. More than a third of respondents said such concerns considerably slow AI initiatives in their organisations, underlining how risk management is shaping the adoption path as much as technical readiness or budgets.

Roman Kilun, Chief Compliance Officer at ABBYY, commented on the findings.

"Businesses are embracing AI, but our research shows that many are still struggling with key questions around accountability, governance and ROI. Senior management needs to do a better job of communicating responsible AI policies across the company. In addition, AI requires organisations to know where data comes from, whether it can be trusted, how it can be used, and where it is stored and processed. As data sovereignty grows in importance, a shift toward machine-readable controls that govern data throughout its lifecycle is required to ensure consistent compliance," said Kilun.

Kilun said the findings also point to a shift from trial projects to broader deployment across businesses.

"Organisations we work with are moving beyond pilots and proofs of concept into execution, and clear frameworks and a shared understanding of responsibility are critical to building confidence in AI. At ABBYY, we're committed to being transparent about our own governance approach and supporting customers on that journey," he said.