Risks and capability

Four AI risk objects open the portfolio. Order carries no rank. Each is read across driver, exposure, vulnerability, capacity, cascade and consequence.

3 min read0 figures1 table11 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

Contents
  1. ·Edition home
  2. ·Executive summary
  3. 1AI is infrastructure, not only software
  4. 2Key findings
  5. 3Systems and sectors at baseline
  6. 4Risks and capability
    1. 4.1 Risk objects
    2. 4.2 Capability ladder
  7. 5Chip supply disruption under grid constraint
  8. 6Lenses and readings
  9. 7The 2027 agenda
  10. 8Systems in depth
  11. AMethodology
  12. BData appendix
  13. CReferences and record
  14. ·Complete edition, for print

4.1Risk objects

Each risk is read across the chain from hazard to consequence, with its status and evidence-quality level.

  1. InferredEQL 3RSK-XT01-MIN-001

    Advanced-chip supply disruption

    Driver
    conflict, disaster or export control in a concentrated supply region
    Exposure
    all AI compute; wider electronics
    Vulnerability
    about 92% of leading-edge capacity in one region (2019)17
    Capacity
    new fabs; stockpiles; diversification
    Cascade
    MIN → AIC → FIN, INF
    Horizon
    1–5 years

  2. ReportedEQL 3RSK-XT01-ENE-002

    Grid and water constraints at data-centre clusters

    Driver
    fast load growth in a few regions
    Exposure
    local grids, water basins, communities
    Vulnerability
    long connection queues; water stress
    Capacity
    grid build-out; efficiency; siting rules
    Cascade
    AIC → ENE, WAT → cities
    Horizon
    1–3 years

  3. InferredEQL 2RSK-XT01-SAF-003

    Failures and misuse of agentic systems

    Driver
    autonomous agents acting in real systems
    Exposure
    finance, cyber, critical services
    Vulnerability
    thin incident data; limited oversight
    Capacity
    evaluations, monitoring, risk frameworks56
    Cascade
    AIC → DAT → FIN, INF
    Horizon
    1–3 years

  4. InferredEQL 3RSK-XT01-LAB-004

    Labour displacement outpacing transition

    Driver
    rapid adoption across tasks
    Exposure
    a quarter to two-fifths of jobs1632
    Vulnerability
    weak retraining and social protection
    Capacity
    skills programmes; social protection
    Cascade
    AIC → labour → FIN, HLT
    Horizon
    3–5 years

4.2Capability ladder

Existence is not readiness. Each of the fifteen capability functions is placed on the ladder by the strongest public evidence for AI at the global level.

Table 4.1Capabilities on the ladder of evidence: listed, exists, demonstrated, tested
FunctionRungLevelStrongest public evidence at baseline
Prevention and mitigationEXISTSrisk-management frameworks published567
Monitoring and observationEXISTSincident databases and monitors2728
Early warningLISTEDno shared warning system for systemic AI failures
Preparedness and planningEXISTSnational AI strategies widespread
Redundancy and reservesLISTEDchip and compute redundancy limited
ResponseEXISTSincident response in firms; cross-border response untested
Continuity of critical servicesEXISTScloud continuity planning; large outages still occur
RecoveryEXISTSstandard IT recovery practice
AdaptationEXISTSskills and education programmes expanding
Social protectionLISTEDlittle AI-specific transition support
Finance and insuranceEXISTSAI liability and insurance markets nascent
Institutional coordinationEXISTSsummits, UN panel and dialogue created1213
Technical infrastructureEXISTScompute concentrated in few providers
WorkforceLISTEDskills shortages widely reported
Research capacityADEQUATElarge research base; safety research small relative to capability3
A capability listed in policy is not a tested one; each rung is the strongest public evidence at baseline.

Notes

Numbers are the edition’s reference numbers, so a note reads the same in every chapter. The full list, with every source and its licence, is in Annex C. Annex C, references

  1. 17Semiconductor Industry Association and Boston Consulting Group (eds.), Strengthening the Global Semiconductor Supply Chain in an Uncertain Era. SIA and BCG, Washington, DC, 2021.back
  2. 5NIST (ed.), Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1. NIST, Gaithersburg, MD, 2023.back
  3. 6NIST (ed.), Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, NIST AI 600-1. NIST, Gaithersburg, MD, 2024.back
  4. 16M. Cazzaniga, F. Jaumotte, L. Li et al., “Gen-AI: Artificial Intelligence and the Future of Work,” IMF Staff Discussion Note SDN/2024/001. IMF, Washington, DC, 2024.back
  5. 32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025.back
  6. 7ISO and IEC (eds.), ISO/IEC 42001:2023, Information technology, Artificial intelligence, Management system. ISO, Geneva, 2023.back
  7. 27Responsible AI Collaborative (ed.), AI Incident Database. incidentdatabase.ai, retrieved 2026-09-30.back
  8. 28OECD (ed.), OECD AI Incidents and Hazards Monitor. OECD.AI, retrieved 2026-09-30.back
  9. 12United Nations General Assembly (ed.), Resolution establishing the Independent International Scientific Panel on Artificial Intelligence and the Global Dialogue on Artificial Intelligence Governance. United Nations, New York, August 2025.back
  10. 13UK Government (ed.), The Bletchley Declaration by Countries Attending the AI Safety Summit, 1–2 November 2023. London, 2023.back
  11. 3Y. Bengio (ed.), International AI Safety Report. Department for Science, Innovation and Technology, London, January 2025.back

Cite this chapter

Valadbaygi, S. (Ed.). (2026). Chapter 4: Risks and capability. In Exponential Technology Report 2026: Artificial Intelligence and Agentic Systems (Baseline Edition 2026, Report No. NR-XTECH-AI-GLOBAL-2026-BASELINE) [Draft under review]. The Global Centre for Risk and Innovation (GCRI).

Every citation form of the whole report, BibTeX and RIS included, is on the edition home. Cite this report

ReportsContribute

Send evidence or a correction on Artificial Intelligence and Agentic Systems.

Reports is the Consortium’s quarterly publication. What you know about this subject can become evidence, a case or a correction in the next edition, credited to you and published under CC BY 4.0 where it is your own work.

GCRIExponential Technology

Never send personal data, sensitive infrastructure detail or anything you have no right to share. A contribution is reviewed before use; sending it confers no authorship or editorial rights, and sponsorship never buys a finding.

The Global Centre for Risk and Innovation

Science · evidence · measurement

Not for profit

Our articles, maps and twins are free to read and free to republish. Readers and partners keep it that way.

Support our work

Who funds us

GCRI is a not-for-profit institution. To ask who funds a piece of work, use the funding questions form.

Independence

Editors and the expert reviewers named on each article decide what we publish. Funders and partners do not review or approve articles before publication.

Corrections policy

Licence

Articles: CC BY 4.0. Figures, maps, twins and data: CC BY 4.0. Other uses: ask through the permissions form.