74%
of organisations considered inaccuracy a relevant AI risk in 2025, up from 60% in 2024; 71% were actively mitigating it

What the chart shows

Inaccuracy was the AI risk most often considered relevant in 2025, by 74% of organisations in the AI Index and McKinsey survey, up from 60% in 2024. Cybersecurity followed at 72%, up from 66%, then regulatory compliance at 63%.

Organisations mitigate fewer risks than they recognise. 71% were actively mitigating inaccuracy and 61% cybersecurity, but for equity and fairness the shares were 30% relevant and 23% mitigated, and for explainability 36% and 29%. Nine of the 13 risks asked about in both years were cited as relevant less often than in 2024.

Key numbers

MeasureValue
Inaccuracy, relevant, 202574.0%
Inaccuracy, relevant, 202460%
Inaccuracy, mitigated, 202570.6%
Cybersecurity, relevant, 202571.8%
Equity and fairness, relevant, 202529.8%
Equity and fairness, mitigated, 202522.6%

How to read it

  • Survey of business leaders by the AI Index and McKinsey & Company in 2024 and 2025, across several regions and industries; it has no respondents from China.
  • The chart shows 2025. The 2024 shares are in the data file; ‘resource misuse’ and ‘autonomous or unintended system actions’ were new options in 2025.
  • Respondents could name several risks.
Source: McKinsey & Company Survey, 2025, via the AI Index Report 2026 (Stanford HAI), figure 3.3.5: AI risks: considered relevant vs. actively mitigated, 2024 vs. 2025. Chart and text by AI Adoption.
Numbers checked against the published data on 20 October 2026.
Download the data (CSV) · Cite this chart