AI, the challenge of governance: companies lack practices and tools to make it reliable

Artificial intelligence, lots of trust but still little governance. Italian companies believe in the potential of technology, but struggle to translate this trust into practices, controls and tools capable of making it truly reliable. In particular, although the 'trust gap' between 'perceived trust' (how much companies trust AI) and 'trustworthiness index' (how much they are actually capable of governing it) has gone from 18.4 points in 2025 to 14.1 in 2026, it remains the widest in Europe. This was revealed by the Data and AI impact Report 2026, a study conducted in April by Idc for Sas on 2,699 decision makers from 28 countries, according to which one of the main critical issues does not concern the adoption of AI so much as the ability to explain its functioning.

Moreover, the more organizations entrust decisions to AI, the more the need to understand its functioning, verify its motivations and demonstrate its reliability grows. However, there are still few companies capable of implementing AI projects. The 'impact index' (the value of the business results actually obtained thanks to investments in AI) grows, in fact, only from 50.2 points in 2025 to 51.7 in 2026 while, thanks to the fragmentation of data, only 18.7% of organizations declare they have achieved a high ROI.

The Report also highlights that data quality, verifiability, responsibility and explainability are factors directly related to the ability to generate measurable business results. Finally, companies that believe and invest in these good practices record significantly higher returns than the average.