WEF Global Cybersecurity Outlook 2026: AI, geopolitics, scams and resilience survey

In two minutes

• AI is named the most significant driver of cyber change in the year ahead by 94% of respondents, and 87% call AI-related vulnerabilities the fastest-growing risk of 2025 (pp. 4-5).
• More organizations now check AI tools before deployment: 64% in 2026, up from 37% in 2025 (p. 4). Roughly one-third still have no process to validate AI security before deployment (p. 21).
• Geopolitics is the top factor in risk mitigation strategies, with 64% accounting for geopolitically motivated attacks (p. 5). Low confidence in national response to major incidents rose to 31% from 26% (p. 6).
• CEOs and CISOs name different top concerns: CEOs put cyber-enabled scams and phishing first in 2026, while CISOs keep ransomware first and supply chain disruption second (p. 8, p. 11).
• Highly resilient organizations differ from insufficiently resilient ones on practical measures, such as assessing AI tool security (83% vs 39%) and having needed skills (78% vs 15%) (table 4, p. 39).
• Supply chain risk is often handled as a checklist: 66% assess supplier maturity, but only 27% run joint incident exercises and 33% map their ecosystem in detail (p. 47).
• OT governance is thin: only 16% of organizations with industrial environments report OT security to the board, and 36% have the CISO responsible for OT (p. 36).
• Inequity persists: small organizations are twice as likely as large ones to report insufficient resilience, and NGOs (37%) and the public sector (23%) trail the private sector (11%) (p. 48).

Key numbers

• 94%: respondents who expect AI to be the most significant driver of cybersecurity change in the year ahead (p. 4).
• 87%: respondents naming AI-related vulnerabilities the fastest-growing cyber risk over 2025 (p. 5, p. 19).
• 37% to 64%: organizations with a process to assess AI tool security before deployment, 2025 to 2026 (p. 4, p. 20).
• 31% (from 26%): respondents with low confidence in their nation's ability to respond to major cyber incidents (p. 6, p. 27).
• 73%: respondents who, or whose network, were affected by cyber-enabled scams and similar crime in the past 12 months (p. 8, p. 30).
• 77%: organizations that have adopted AI for cybersecurity, mainly phishing detection (52%) (p. 22).
• 65% (from 54%): large companies by revenue naming third-party and supply chain vulnerabilities as their greatest resilience challenge (p. 45).
• 804 participants, 92 countries: qualified survey sample (873 responses received, 316 CISOs, 105 CEOs), run 25 Aug to 1 Oct 2025 (p. 57).

What to do with it

• Check whether every AI tool goes through a security assessment before deployment, and whether reviews repeat. The report links this to resilience (table 4, p. 39; figure 34, p. 41).
• Review identity and privilege controls for AI agents. Box 1 (p. 21) says agents can gather excess privileges or be manipulated, and calls for continuous verification, audit trails and zero-trust principles.
• Move supply chain work beyond supplier questionnaires: map the ecosystem, run joint incident exercises and bring security into procurement, as the more resilient organizations do (figure 35, p. 42; p. 47).
• Take OT security to the board and clarify who owns it, given the low reporting and CISO ownership rates (p. 36).
• Use the CEO and CISO risk lists (table 1, p. 8) to align board conversations on scams and phishing, AI vulnerabilities, ransomware and supply chain.
• Use the report's incident cases (Jaguar Land Rover costs, p. 44) to support cyber risk quantification and scenario building, which the report recommends.

Read the full document if

• you need survey benchmarks on AI security practice, supply chain measures or resilience traits to compare with your own programme.
• you brief a board or CEO and want data on how CEO and CISO priorities differ.
• you need the regional and sector breakdowns on confidence, resilience and skills shortages (figures 20, 40, 41, 44).

What it relies on

The main source is the Global Cybersecurity Outlook survey, run 25 August to 1 October 2025. It drew 873 responses from 99 countries, of which 804 from 92 countries qualified. Other inputs are 52 focus-group participants, workshops with Global Future Council members and 21 CISO Community executives, and short polls at an October 2025 meeting (p. 57). The report also cites news, government and agency sources in its endnotes, plus earlier Forum papers. Accenture is the named collaborator, and its Global Cybersecurity Lead co-signs the foreword. The Centre for Cybersecurity partners include Accenture, Fortinet, IBM and Palo Alto Networks (p. 59). The report says it used genAI to support research and survey translation (p. 57).

Our read

A survey-based reference with many breakdowns by region, sector, size and resilience level. It is useful for benchmarking and for board material. The sample is self-selected and CEO subgroups are small, so treat segment figures as indicative. A few text figures differ from the charts, for example on geopolitical strategy change (p. 24 and figure 16).