Sunday, 18 October 2026

Follow on Google

diarioempresas

IBEX 3519.326,30▲ +0,77%EuroStoxx 506204,23▲ +0,58%S&P 5007659,44▼ -1,09%€/$1,1242▲ +0,19%Brent109,86▲ +3,72%Bitcoin73.058▲ +0,89%
Breaking
Nine Andalusian startups enter the Top 101 most innovative in Spain49% of large companies have not updated their AI governancePublic stakes in the Ibex, with €50 billion at stake on 29-NCésar Vargas leads Pascual's transformation after nearly a year at the helmWhat is a business angel: who they are, how much they invest, and what they bring to a startupSpain adds 3.44 million employed since 2018 and the minimum wage rises by 65.9%
Companies

49% of large companies have not updated their AI governance

85% of executives acknowledge some AI systems operate without human oversight, according to an EY survey on agentic AI risks.

Beatriz Lorenzo Aguirre
Beatriz Lorenzo Aguirre
· 2 min read
AIIllustrative image: modern office corridor with glass converging to a vanishing point, illuminated by natural light
Illustrative image: modern office corridor with glass converging to a vanishing point, illuminated by natural light

91% of large companies are already using agentic AI in pilots or full deployments, but 49% have not updated their governance framework. This is reported by the AI Risk and Governance survey from EY.

The 85% of senior executives surveyed in the AI Risk and Governance study by EY acknowledge that at least some of their AI systems perform actions without real-time human involvement. The survey, published in September 2026, includes responses from 202 senior AI executives in companies with at least US$1 billion in annual revenue.

The information, advanced by ecosistemastartup.com, highlights the issue of the gap between technology use and controls. According to the same study, 26% cannot detect unauthorized AI agents operating within the organization, and 36% experienced a materially negative incident in the past year, such as data loss, financial damage, or operational disruptions.

A year earlier, the Responsible AI Pulse from EY, published in June 2025, indicated that 72% of senior executives had integrated AI into most or all of their initiatives. Only about a third had controls for all components of a responsible use model.

A study by Deloitte in January 2025 estimated that 69% of surveyed companies believed a complete governance strategy would take more than a year to implement.

The Cost of a Data Breach Report 2025 from IBM measured that 63% of organizations that suffered data breaches did not have AI governance policies or were developing them. Companies with high levels of shadow AI, meaning the use of applications not approved by the company, recorded average breach costs USD 670,000 higher than those with little or no uncontrolled use.

In the report The State of AI in 2025 from McKinsey, 30% of organizations using AI reported at least one negative consequence linked to inaccuracies.

Regarding agentic AI, EY noted that 76% of companies were using or planning to incorporate it in the next 12 months, but only 56% claimed to be aware of its risks. Finance, human resources, sales, and customer service are among the areas with the highest exposure.

In the 2026 study, 49% of respondents admitted they had not yet updated their governance framework to include the specific risks of agents.

Beatriz Lorenzo Aguirre

Written by

Beatriz Lorenzo Aguirre

Redactora

Periodismo económico por la Carlos III y lectora compulsiva de cuentas anuales. Cafés a destajo, alergia a las notas de prensa vacías y memoria para los ERE; en Diario Empresas escribe de empresas y empleo.

You may also like