Deloitte: 40% of companies now let AI execute decisions without human approval
Direct answer
On October 29, 2026, Deloitte published State of AI in the Enterprise, its annual report based on a survey of 3,500 director- to C-suite-level leaders across 25 countries and six industries, with fieldwork in May and June 2026. The headline finding for AI buyers: 40% of surveyed companies allow AI systems to execute decisions or actions without human approval, at least to a moderate degree, while 23% say accountability is unclear when AI produces a negative outcome. Autonomy has reached operations; the governance around it is still catching up.
The numbers that matter
- 81% report faster decisions: most surveyed companies say AI sped up decision-making and execution, the report's most widespread benefit.
- 28% report AI revenue gains: the figure rose from 20% in the January 2026 report, a sign that financial value is consolidating beyond pilots.
- 25% redesign work at depth: only one in four companies uses AI to create new products, reengineer core processes or reinvent its business model. The rest accelerate what they already did.
- 40% with moderate-or-greater autonomous execution: agents approving, routing, purchasing or answering on their own, within limits each company defines.
- 23% with unclear accountability: nearly one in four companies admits it is undefined who answers when an AI outcome goes wrong.
What to do with this report in your business
- Define approval tiers before granting autonomy: low-amount, low-risk routine work can run on its own; money, customers and external commitments need human approval. Write it in a table, with amounts and cases.
- Name an owner for every agent: each automation has a human owner who reviews its mistakes weekly. Diffuse accountability is the risk the report's 23% describes.
- Require a complete log: every AI-executed decision leaves a trail: which data it used, which rule it applied and who could stop it. Without a log, a dispute becomes impossible to reconstruct.
- Start with a high-volume, low-risk process: classifying, routing and answering repetitive work delivers the 81% speed gain without exposing cash. Autonomy over money comes later, with tested limits.
- Measure revenue, not just activity: the 28% reporting gains measure financial outcomes. Count closed cases, time saved and attributable sales with the same discipline as any investment.
Note: this article summarizes Deloitte's October 29, 2026 press release and Daily AI Brief and AI Agent Update coverage of State of AI in the Enterprise. The figures cover companies already using AI in daily operations, per the report's methodology, and describe what respondents reported.
Want AI that works for your business?
Message us on WhatsApp and let's talk about automation and AI agents for your business.
WhatsApp inicIARelated articles
- AI spending nears $400B but only 23% scale agents
- How to tell if your AI is working: the 5 KPIs that matter
- What to do when your AI agent makes a mistake
Frequently asked questions
What is Deloitte's State of AI in the Enterprise?
The Deloitte AI Institute's annual report on how companies use AI. The October 2026 edition surveyed 3,500 director- to C-suite-level leaders across 25 countries and six industries, in May and June 2026.
What does the 40% without-approval figure mean?
That 40% of surveyed companies allow AI to execute decisions or actions without human approval, at least to a moderate degree. The report does not say every decision is autonomous: each company defines its limits.
Why does the 23% unclear-accountability figure matter?
Because one in four companies admits it is undefined who answers when AI produces a negative outcome. That gap is the first governance risk to close before expanding autonomy.
What should a small business do with this report?
Start with one repetitive, low-risk process, define approval tiers in writing, name an owner per automation, and measure revenue and time saved, not just activity.
