Skip to content
22decisions home

Research brief · X-02

The CISO as AI Gatekeeper: What It Takes to Get Approved

In editorial reviewPlanned publish date: 26 November 2026
Theme
Cross-industry
For
CISO

Summary

A cross-industry brief on how enterprise AI decisions are changing, drawing on the 22decisions decision-makers panel, hiring data, and published research, with implications for technology leaders and the vendors that sell to them.

The short answer

The CISO is the gatekeeper for which AI tools an organisation can use, and with what data. In the 22decisions decision-rights map, the CISO decides or approves 6 of the 22 AI decisions, and across large organisations 43% of technology staff say their IT or security teams have rules that block AI agent tools.

Key questions

  • Which AI decisions does the CISO control?
  • How cautious are CISOs about AI, by industry?
  • What does a CISO need to see before approving an AI tool?
  • How should AI vendors prepare for security review?

Headline finding

CISOs are the most cautious role in AI decisions: in our model, 77% of banking CISOs and 83% in automotive and manufacturing are Stewards or Sceptics.

Calibrated modelCalibrated model.

What the full report covers

  • Who is involved in this decision, who decides, and who can say no
  • How AI leaders approach it differently from laggards
  • What technology leaders should do next
  • What it means for the technology and services partners they buy from

Full report: 8 to 10 pages, PDF.

Evidence basis

  • Calibrated model
  • Cited
  • Measured

How we label evidence is explained in our methodology.

Frequently asked questions

Who approves new AI tools in large organisations?
Usually the CISO decides, with risk, compliance, and legal able to veto (Calibrated model).
How common are security rules that block AI tools?
43% of technology staff at large organisations say IT or security rules block AI agent tools (Measured: Stack Overflow Developer Survey 2025).
How can vendors speed up security approval?
Provide security architecture, data-handling details, and controls before the pilot, not after.

Get it first

This brief is planned for 26 November 2026. Request early access and we'll send it to you as soon as it is ready.

The same decision in other industries

More from Banking & Financial Services

Get our research in your inbox

One considered briefing a month on how organisations decide on, buy and adopt AI. Unsubscribe at any time.