UK business leaders remain divided over who should be responsible when artificial intelligence produces harmful or incorrect results, as companies struggle to build governance systems quickly enough to keep pace with adoption.

More than a third of senior managers at businesses using AI believe responsibility should be shared between the company deploying the technology and its supplier, according to research commissioned by AI software company ABBYY.

The survey of 200 senior managers at UK businesses with more than 100 employees found 35 per cent favoured shared responsibility. Some 27 per cent said the organisation using the AI should be primarily responsible, while 19 per cent placed responsibility with the technology vendor. Nine per cent blamed the end user and 10 per cent said regulators should take the lead.

The uncertainty comes as 57 per cent of respondents said AI was being introduced faster than their organisation could effectively govern it.

Confidence in the technology also remains limited. Only 26 per cent said they completely trusted AI to produce accurate outputs, while 22 per cent completely trusted systems to operate without creating unacceptable risks.

Data governance is proving another obstacle. Seven in 10 respondents said requirements governing the accuracy, security and compliance of data were slowing AI deployment. Integration with existing technology was cited by 20 per cent as their biggest obstacle to generating greater returns from AI, followed by data quality at 17 per cent and regulatory uncertainty at 15 per cent.

Despite these concerns, UK companies continue to increase spending. Respondents reported average investment of £889,000 in AI tools during the past 12 months, up 6 per cent from £840,000 the previous year.

Most companies surveyed already had some governance structures in place, with 81 per cent reporting a formal framework. But only 53 per cent had an AI incident reporting process, 44 per cent an incident response plan and 29 per cent a mechanism to shut down an AI system.

Roman Kilun, chief compliance officer at ABBYY, said companies were "still struggling with key questions around accountability, governance and ROI".

"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," he said.

The findings suggest governance could ultimately accelerate rather than constrain adoption. More than three-quarters of respondents said governance had made their AI initiatives more successful, while 63 per cent said it made moving projects beyond the pilot stage easier.