Global survey finds that most organizations are scaling AI without the governance, measurement, or accountability needed to prove it works.

Key Takeaways

• 78% say they can’t pass an AI governance audit: Most organizations lack confidence that they could produce documented evidence of AI controls and outcomes within 90 days.
• Governance, not technology, drives AI underperformance: 46% cite governance or compliance gaps as the leading cause.
• Board oversight gaps persist: While 75% of boards approved AI investments, 48% have not set governance expectations, and 46% have not integrated AI risk into oversight processes.
• AI strategy is crucial but not implemented: 51% say strategy drives AI ROI, yet only 22% have a fully implemented enterprise AI strategy.
• Workforce readiness is low and under-resourced: Only 12% say their workforce is fully ready, while 97% report adoption challenges and 34% cite underfunded training.
• AI risk preparedness is limited: 74% allow AI access to internal systems, but only 20% have tested incident response plans.

Grant Thornton’s 2026 AI Impact Survey measures how organizations deploy, govern, and gain from AI. The survey conducted 950 live interviews of senior leaders across 10 industries, including finance, manufacturing, technology, and private equity. Respondents included C-suite executives and functional leaders across operations, finance, and IT. 

Executives doubt their governance

The survey found that 78% of executives doubt their companies could pass an independent AI governance audit within 90 days.

Executive boards are committed to AI, just not governance

Three in four boards approved major AI investments, but 48% have not set governance expectations, and 46% have not integrated AI risk into board oversight processes.

The gap extends into performance

When asked what factors contributed to AI underperformance, only 8% of respondents said AI is not underperforming, meaning the other 92% reported some level of failure or unmet expectations. 46% cited governance or compliance barriers as the primary driver, compared to 15% citing technology limitations. 

The report calls this disconnect the “AI proof gap,” where organizations deploy AI systems but cannot explain how they work, measure their outputs, or defend their decisions under scrutiny.

Executives know strategy is key; few are doing it

The findings also show that most organizations lack a defined strategy for AI. While 51% of executives identified strategy as the biggest driver of AI return on investment (ROI), only 22% have a fully developed and implemented enterprise AI strategy, and 53% are still building one.

The workers aren’t ready for AI

Just 12% of leaders said their workforce is fully ready to adopt AI, while 81% said their workforce is only partially ready. At the same time, 34% identified training as the most underfunded area of AI investment, and 97% reported adoption challenges across their organization. 

Lower levels responsible for AI lack resources and direction

Frontline employees (37%) and middle managers (30%) account for two-thirds of where organizations say support is most needed, indicating that those responsible for implementing AI lack resources and guidance. 

Companies are engaging autonomous AI; few are prepared if something goes wrong

74% of organizations are allowing autonomous AI to access internal systems and data. However, only 20% have tested an AI-specific incident response plan. Further, 48% created a response plan, yet never tested it. 

AI maturity = confidence in governance 

Among organizations with fully integrated AI systems, 74% said they are very confident they could pass an audit, compared to 7% of those still piloting AI. 

Full AI integration reaps tangible rewards

Organizations with fully integrated AI reported 81% efficiency gains, 58% revenue growth, 64% higher output quality, and 59% accelerated innovation. By contrast, organizations still piloting AI reported 15% revenue impact and 24% innovation gains.

Further, Grant Thornton does not define regulatory requirements for AI governance or specify audit standards, nor does it assess compliance with specific laws. Instead, it measures executive confidence in meeting independent audit expectations and producing supporting evidence.