Companies worldwide are deploying artificial intelligence (AI) systems far faster than they’re managing the associated risks, according to research from the Thomson Reuters Foundation’s AI Company Data Initiative (AICDI).
The initiative, developed in partnership with UNESCO, analyzed AI governance disclosures from 1,000 companies across 13 sectors, creating what is currently the world’s largest dataset on corporate AI oversight practices.
The research underscores that most companies remain in early stages of understanding AI’s operational and environmental footprint, highlighting an urgent need for more comprehensive oversight frameworks that address both governance risks and real-world energy impacts.
The findings expose a striking disconnect between AI adoption and responsible governance. Most alarmingly, 97% of companies failed to assess the environmental impact of their AI systems, including energy consumption and carbon footprints — a critical oversight as concerns mount about AI’s contribution to rising electricity demand and emissions.
While companies frequently describe their AI systems as “ethical,” “trustworthy,” or “secure,” nearly none connect these deployments to their broader climate commitments or environmental responsibilities.
Donatas Karčiauskas, CEO of Exergio, a firm specializing in energy efficiency tools for real estate, argues companies need more sophisticated evaluation frameworks.
“Many people assume AI will always waste energy, so they never stop to ask about its environmental impact. But that’s not true,” he said, explaining that advanced building management systems can reduce heating and cooling demand.
The governance gaps extend beyond environmental concerns. While 76% of companies with AI strategies claim oversight at the management level, only 41% make their policies accessible to employees or require policy acknowledgment.
Karčiauskas calls this a “clear warning sign” about flawed AI impact assessment. “If you don’t watch what AI is doing in real time, you’re guessing whether it helps or harms your goals,” he said, calling for operational data rather than mere paperwork.
Regional differences emerged in the study, with EMEA companies (Europe, Middle East, and Africa) leading in published AI strategies at 53%, largely driven by the EU AI Act. However, even in Europe, environmental impacts remain largely absent from governance disclosures.
“Europe is ahead on regulation, but even here the energy footprint of AI is mostly absent from the discussion,” Karčiauskas said. He encouraged operational transparency, including power consumption metrics, to become standard governance practice.
