UN chief urges AI companies to 'come clean' about the pollution they generate

On June 23, 2026, UN Secretary-General António Guterres formally proposed the 'AI Environmental Transparency Initiative' at London Climate Action Week, calling on global AI companies to 'come clean' about the pollution generated by their operations. Guterres pointed out that as AI models continue to scale up, the energy, water, and electronic waste required for training and inference are increasing dramatically, yet the public and regulators know little about it.

The initiative requires companies to disclose three key indicators: energy consumption (including electricity sources and carbon emissions), water usage (primarily for data center cooling), and electronic waste (hardware replacement cycles and recycling rates). Guterres emphasized that without transparency, it is impossible to measure the true environmental cost of AI or formulate effective emission reduction policies.

Currently, the initiative is voluntary and lacks legal force. Whether companies participate depends entirely on their own will. Guterres acknowledged that it is difficult to establish a globally unified mandatory disclosure standard in the short term, but hopes the initiative will drive industry consensus and lay the groundwork for potential future regulation.

However, the main obstacle for companies is data confidentiality and trade secrets. AI companies' energy efficiency, data center layouts, and supply chain information are often considered core competitive advantages; disclosing such data could weaken their competitive edge. Additionally, the lack of unified accounting methods (e.g., how to calculate indirect emissions) adds to the difficulty of disclosure.

Some companies have already begun voluntarily disclosing partial data. For example, Google and Microsoft have committed to achieving carbon neutrality by 2030 and publish regular environmental reports. But critics point out that these reports often lack detail, and data across different companies are not comparable.

After the initiative was proposed, some AI companies welcomed it, believing transparency helps enhance industry image and attract green investment. However, other companies are taking a wait-and-see approach, concerned about increased operational costs and legal risks. Non-governmental organizations are calling on the UN to push for stronger measures, including incorporating environmental disclosure into international trade agreements.

In the long run, the initiative may drive the AI industry toward more sustainable development. If major companies take the lead in disclosing data, it will create a demonstration effect and encourage upstream and downstream supply chains to follow. At the same time, investors and consumers will gain more information, thereby influencing market choices.

But uncertainties remain: Can the voluntary framework generate enough pressure? Will companies selectively disclose favorable data? Without third-party audits, how reliable is the data? These questions require ongoing observation.