A visualization published by Visual Capitalist, based on a McKinsey survey of 1,753 respondents conducted between June and July 2025, found that inaccuracy is the most commonly reported danger of AI in the workplace. Nearly a third of companies, 30%, said they experienced at least one negative consequence tied to AI producing inaccurate output.

Explainability, meaning the difficulty of understanding how an AI system reached a decision, ranked second at 14%, followed by personal and individual privacy violations at 11% and cybersecurity incidents at 10%. Regulatory compliance failures and intellectual property infringement tied at 8% each, while unauthorized or unintended actions by AI systems and equity and fairness problems each affected 7% of companies. Workforce displacement, often cited as AI's most visible risk, ranked lowest among the categories measured at 6%.

More than half of the businesses surveyed reported experiencing at least one negative consequence from AI overall. The report noted that inaccuracy-driven errors carry consequences beyond the initial mistake, undermining trust in AI systems, causing operational inefficiencies, and in some cases prompting insurers to add exclusions for AI-related losses to corporate liability policies. Visual Capitalist reported that 54% of companies are actively working to reduce AI inaccuracy, generally by keeping a human reviewer in the loop before AI-generated output reaches use.

Source: Visual Capitalist - https://www.visualcapitalist.com/sp/ter02-the-dangers-of-ai-top-risks-companies-face/