CD Consulting R&D

CD Consulting R&D · KB-RISK · entry 1 companion · compiled 2026-09-07 from the seven public WEF Global Risks Reports · published 7 September 2026

Across seven editions the Global Risks Report has rebuilt what it means by a technological risk. In 2020 the category was about systems failing — cyberattacks, data theft, infrastructure breakdown. By 2026 it is about truth failing, work disappearing and control slipping: misinformation second over two years, AI fifth over ten. This page sets out the movement, the break year, and the one number that defines the AI risk.

01Read the ranks with care

The WEF changed its central question twice in seven years. In 2020–2021 risks were scored on likelihood and impact over ten years. In 2021–2022 a "Global Risks Horizon" question asked when a risk becomes a critical threat. Only from 2023 does a single question run consistently: estimate the likely impact (severity) over a two-year and a ten-year period.

The list changed too — 30 risks in 2020, 37 in 2022, 33 in 2026 — and so did the names. Cyber insecurity became Cyber espionage and warfare in 2025 and reverted in 2026. Technological power concentration was dropped after 2024. Online harms was added in 2025. And there is no stand-alone AI risk at all before the 2024 edition.

02Seven years in one column

• 2020Technology as an infrastructure and governance problem. Data fraud (#6) and cyberattacks (#7) in the likelihood top 10; information infrastructure breakdown (#6) and cyberattacks (#8) by impact. AI appears in the Wild Wide Web chapter as an upside — "expected to boost global growth by 14% by 2030."
• 2021The pandemic reframes technology as dependency and divide. Digital power concentration (#6) and digital inequality (#7) by likelihood. In the new horizon question, Adverse tech advances ranks #4 among 5-to-10-year "existential threats" — the long-horizon anxiety that AI will later carry, without the name.
• 2022The trough — and the WEF's own warning. No technological risk in the ten-year top 10. The report calls this "a possible blind spot."
• 2023Cyber promoted, AI absent. Widespread cybercrime and cyber insecurity enters both top 10s at #8. The survey closed in autumn 2022, before generative AI reached the public; there are thirteen mentions of AI in the whole document.
• 2024The break. Misinformation takes #1 over two years — the first technological risk ever to top the ranking. Adverse outcomes of AI technologies enters the list and lands at #6 over ten years. And "AI-generated misinformation and disinformation" is named the #2 current crisis of the year, by 53% of respondents.
• 2025Consolidation, and a warning against complacency. Misinformation holds #1; AI holds #6 over ten years but slips to #31 over two. Under the heading "Technological risks — still under the radar," the report writes that "complacency around the risks of such technologies should be avoided."
• 2026AI arrives in the present tense. Geoeconomic confrontation takes #1, pushing misinformation to #2 — but misinformation rises to #4 over ten years, its best long-term placement. AI is a new entrant to the current-risk top 10 at #8, and the report names it "the risk with the largest rise in ranking over time."

03The comparable series, 2023–2026

The series stabilises after 2024: two technological risks in the two-year top 10 and three in the ten-year top 10, every year. What moves is composition, not volume — misinformation deepens long term, censorship and surveillance is the only technological risk climbing steadily on the short horizon, and concentration is reclassified out of the category into an economic risk, Concentration of strategic resources and technologies.

04The AI gap, measured

Four readings of the gap

1. The panel is right and the risk is genuinely deferred. Labour displacement at scale, military integration and information ubiquity are decade-scale phenomena.
2. The panel is systematically late — and says so. The WEF has flagged this twice: "a possible blind spot" on long-range technological risk (2022) and "complacency … should be avoided" (2025). The corpus contains a documented lag: state-based armed conflict "was overlooked as a leading two-year risk two years ago" before becoming the #1 current risk.
3. The instrument favours event-shaped risks. AI's short-term harms are attributed elsewhere — a laid-off worker to Lack of economic opportunity, a deepfake to Misinformation, an AI-enabled intrusion to Cyber insecurity. Both of the latter sit in the 2026 two-year top 10, and both are described in that report as AI-amplified.
4. The instruments disagree, and the disagreement is the signal. The expert panel puts AI at #30 over two years. The current-risk question puts it at #8 for 2026. The Executive Opinion Survey has three countries ranking it their single most important national risk and twenty placing it top-five. The two instruments closest to operating businesses are the more alarmed.

05What the AI risk is now said to be

The content of the argument changed as decisively as the rank. The 2024 chapter, AI in charge, is a governance and market-structure case: concentration in a supply chain that "favors a few companies and countries", the black-box problem compounded by the pacing problem, and the observation that "the precautionary principle has largely not been applied in the development of AI, as regulators erred on the side of innovation."

The 2026 chapter, AI at large, is a social-consequence case. Its three headline claims:

In a worst-case scenario for labour markets, increases in both productivity and unemployment could drive permanently K-shaped economies.

The potential for creativity, learning and leisure could give way to loss of purpose, meaning and contribution to society, coupled with erosion of alignment around objective facts.

The rising range of military use cases for AI will come with commensurate risks, in the worst case leading to rapid and perhaps unintentional escalation of conflicts.

Global Risks Report 2026, Section 2.7 — AI at large

The report explains its own change of emphasis: AI "has shifted from a frontier technology to a systemic force shaping economies, societies and security," with the global market projected from roughly $280 billion in 2024 to $3.5 trillion by 2033.

Jobless productivity

A benign estimate — 170 million roles created against 92 million displaced, a net gain of 78 million by 2030 — is set against a severe one: up to half of US entry-level white-collar jobs eliminated within five years, unemployment at 10–20%. The chapter coins the image of a "white-collar rust belt" in cities that are today knowledge and services hubs, "generating a powerful, angry, political force." The transmission mechanism is named: Inequality is the most interconnected risk in the survey for the second year running — "labour displacement feeds inequality, which drives societal polarization."

Purpose in drift

New material with no precedent in earlier editions: a risk-register entry about meaning. Loss of occupational identity, a possible "lost generation" subsisting on basic income, dependency on AI as a "second brain", AI companions increasing loneliness, and the erosion of critical thinking in education systems — alongside second-order physical effects, from generative models consuming "up to 4,600 times more energy than traditional software" to water stress around data-centre build-outs.

Military misuse or mistakes

The most concrete escalation scenario in the corpus. As militaries embed AI in "intelligence, surveillance, logistics, and command functions," the risk shifts "from tactical to systemic." Crises "that once unfolded over days or hours could instead escalate in seconds," producing "a conflict started by technical error rather than strategic intent" — with countries already "starting to implement AI tools for managing nuclear weapons stockpiles and in some areas of nuclear weapons command, control, and communications."

Major powers are rushing to integrate AI across military domains, each fearing strategic disadvantage if rivals move first. This dynamic incentivizes rapid deployment over rigorous safety testing, increasing the probability of failures precisely where consequences are most severe … Each country's pursuit of security may, collectively, produce a more dangerous world.

Global Risks Report 2026, Section 2.7

06Ten findings

1. The technological category was remade, not extended, between 2022 and 2024 — four of six risks replaced, and the vocabulary moved from infrastructure and access to content, control and cognition.
2. 2022 is the trough and the tell: no technological risk in the ten-year top 10, and the WEF flags "a possible blind spot" two years before AI enters at #6.
3. 2024 is the break: misinformation at #1 over two years, AI entering at #6 over ten, and AI-generated misinformation the #2 current crisis at 53%.
4. The post-2024 series is stable in volume, unstable in composition — two technological risks in the two-year top 10 and three in the ten-year top 10, every year.
5. AI's signature is the horizon gap, the largest in the 2026 survey: #30 versus #5, severity 3.50 versus 5.28.
6. 2026 is the first year AI is a present-tense concern — a new entrant at #8 in the current-risk landscape, and a top-five national risk in twenty economies.
7. The AI argument itself changed from governance and market structure (2024) to social consequence (2026).
8. Information integrity is the through-line — the only technology theme in the two-year top 10 in every edition since 2024, and the route by which AI risk is expected to reach societies first.
9. Frontier technology other than AI has not moved at all — last or near-last among technological risks in both horizons for four consecutive editions, including in years it received dedicated chapters. The panel's long-horizon concern is specific to AI.
10. Three instruments inside the 2026 report disagree about AI's urgency, and the two closest to operating businesses are the more alarmed. That disagreement is more informative than any single rank.