Corporate governance gaps flagged

Companies are rolling out artificial intelligence faster than they are showing how it is governed, according to a 2026 report published by UNESCO and the Thomson Reuters Foundation. (unesco.org) The report draws on nearly 3,000 companies across 11 sectors and 6 regions, using about 100,000 public data points collected through the AI Company Data Initiative. It was released on March 31, 2026, as the first major findings from that UNESCO-backed disclosure effort. (trust.org 1) (trust.org 2) Nearly half of the companies sampled said publicly that they have an artificial intelligence strategy, but only just over one in ten publicly committed to a recognized governance framework or standard. Less than a third disclosed a dedicated team or resource for artificial intelligence governance. (trust.org) Worker safeguards were thinner still. Just over one in ten companies reported policies to reduce harm to workers from artificial intelligence systems, and only about a third said they offered any artificial intelligence-related training to employees. (trust.org) The same gap shows up in risk checks. Roughly one in ten companies said a human oversees artificial intelligence systems, while only 7% reported assessing the human rights impact of those systems. (trust.org) The dataset also found companies often write broad principles without showing how they work in practice. Thomson Reuters Foundation said 71% of companies with an artificial intelligence strategy mention ideas such as “ethical,” “safe,” or “trustworthy,” but 68% did not adequately assess wider social effects and 97% did not consider environmental impact in deployment decisions. (thomsonreuters.com) That disclosure gap is becoming a market issue as well as a policy one. The initiative says companies are being asked to map artificial intelligence use across products, operations and services so boards, investors and regulators can compare controls, accountability and audit readiness. (aicdi.trust.org) (trust.org) UNESCO’s benchmark is its Recommendation on the Ethics of Artificial Intelligence, adopted by 194 member states on November 23, 2021. The standard calls on private companies to build due diligence and supervision mechanisms and be accountable for artificial intelligence impacts on human rights, the rule of law and inclusive societies. (unesco.org) (trust.org) Companies that do cite an outside framework most often point to the European Union’s Artificial Intelligence Act, ahead of the United States National Institute of Standards and Technology Artificial Intelligence Risk Management Framework and International Organization for Standardization standards. The UNESCO report said more than half of companies that disclosed adherence to a framework cited the European Union law, including many operating outside Europe. (unesco.org) UNESCO and the Thomson Reuters Foundation launched the AI Company Data Initiative in October 2024 and opened the survey for company responses in July 2025. The report’s closing message is that artificial intelligence governance is moving from voluntary language toward evidence that buyers, investors and regulators can check. (trust.org) (aicdi.trust.org)