Corporate AI governance gaps highlighted in UNESCO report
Corporate AI adoption is accelerating, but AI governance, impact assessment, and worker protections remain uneven, a UNESCO report says.
UNESCO and the Thomson Reuters Foundation have published ‘Responsible AI in practice: 2025 global insights from the AI Company Data Initiative‘, presenting findings from what the report describes as the largest global dataset of corporate responsible AI disclosures.
The report analyses 2,972 companies across 11 sectors and multiple regions using publicly available disclosures and company survey responses collected through the AI Company Data Initiative.
The report says AI is being embedded across companies’ products, services, and internal operations faster than governance and disclosure are developing. It states that 43.7% of companies publicly communicate having an AI strategy or guidelines, but only 13% publicly claim adherence to a formal AI governance framework.
Among those that do cite a framework, 53% refer to the EU AI Act, while the report says 43.6% cite ‘other’ frameworks, which it presents as weakening comparability across the wider AI governance ecosystem.
The publication also says many companies describe AI governance in conceptual terms while providing less evidence on operational controls, accountability pathways, monitoring, and remediation. It states that 40% report board- or committee-level oversight on AI, and 12.4% report having a policy to ensure a human oversees AI systems.
At the same time, the publication says 72% of companies do not report conducting any AI-related impact assessment. Of those that do, 11% report environmental impact assessments and 7% report human rights impact assessments. The key statistics on page 10 visually present these findings.
Regarding labour impacts, the report says companies do not provide adequate protection for workers as AI reshapes jobs. It states that while 31% of companies claim to have AI training programmes, only 12% offered structured training with comprehensive coverage. It also argues that effective worker protection requires stronger evidence of reskilling, retraining, redeployment, transition support, and access to remedy where AI affects workers’ rights.
Why does it matter?
The report further states that ethical issues, including human rights and environmental impacts, are being sidelined in AI governance and risk management, while transparency regarding training data, third-party systems, and user rights remains uneven. It presents the AI Company Data Initiative as a tool to help companies assess their governance practices against UNESCO’s Recommendation on the Ethics of AI and to give investors more comparable information on how AI is governed in practice.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!