Company secretaries have revealed a lack of confidence in artificial intelligence, with three-quarters in a survey saying they are concerned about its “impact” on accuracy.

The Chartered Governance Institute (CGI) of UK and Ireland published a report showing 74% of the more than 600 company secretaries polled worry about AI’s precision when it comes to helping with corporate disclosures.

More than a third, 37%, say their biggest challenge is board understanding of AI technology.

Peter Swabey, director of policy at CGI, says AI is already used informally, and without oversight in many companies, including for governance processes.

“While tools such as Copilot can offer real efficiency gains, our research shows that governance professionals are deeply concerned about the risks to accuracy, ethics and trust.

“This report is a wake-up call for boards: they need to develop clear strategies, invest in training and ensure AI use aligns with sound governance principles.”

The poll comes at a time when public discussion has focused on generative AI’s ability to produce accurate output.

Core concerns

A report from tech giant Apple, which received widespread coverage, said its own research had found that leading AI models—known as large reasoning models—experience “complete accuracy collapse”—when faced with problems that require certain levels of exactness.

Elsewhere, research shows that companies could be spending a global total of $36bn on AI governance advice and consultants in ten years’ time.

Meanwhile, recent research from consultancy EY reveals that half of companies believe it is “challenging” to develop governance for current technologies.

The CGI report finds that 43% of company secretaries believe the most challenging aspect for boards is understanding the technology, while 36% say risk management is the greatest issue. Just 11% say their top concern is ethical considerations.

With 84% of governance professionals saying AI is currently being used, only 12% say it has been deployed.

Only 8% of those questioned could say they were “not concerned at all” about the impact of AI on the precision of disclosures.

CGI concludes: “Generative AI represents a powerful tool for enhancing corporate reporting and broader business practices, offering efficiency gains and new capabilities for handling complexity.

“However, its adoption must be approached thoughtfully, balancing innovation with robust governance and transparency.

“Human expertise remains indispensable, both to validate AI-generated content and to preserve the authenticity and strategic value of reporting.”

The UK’s Institute of Directors (IoD) has unearthed AI worries too. In April, the IoD company leaders that used AI were achieving better data insights and analytics but half of the 600 directors polled said “limited experience” among managers and board members, as well as a “lack of trust” in AI outcomes, were their biggest concerns.

Concerns about AI are evolving. Despite best efforts by the tech providers, the original hype is subsiding. In its place are ongoing concerns about the reliability of genAI outputs and its ability to tackle complex problems. AI remains short of being ‘a technology for all reasons’.