Petre Roman calls for global AI treaty and stricter transparency rules

Former Romanian Prime Minister Petre Roman said AI now moves faster than governments can regulate and urged a global treaty to limit risks without slowing useful innovation. In a podcast recorded at the Club de Madrid summit in Madrid, he also warned about disinformation, weak transparency and the environmental strain from data centers.

Why it matters: - Petre Roman said AI is advancing faster than national regulators can keep up, raising risks for democracy, critical infrastructure and public trust. - Roman argued that only international coordination can set guardrails broad enough to cover systems built and deployed across borders. - Roman also tied AI governance to environmental pressure, especially the growing water demand from data centers.

What happened: - Former Romanian Prime Minister Petre Roman spoke with host Sanjay Puri on the RegulatingAI Podcast. - The conversation was recorded live at the Club de Madrid Annual Policy Summit in Madrid. - Roman called for a global treaty on AI and said regulation should protect society without blocking beneficial innovation.

The details: - Roman said frontier AI models can already identify vulnerabilities in critical infrastructure. - He argued that voluntary limits from some companies are temporary because comparable systems will emerge elsewhere. - Roman compared the AI race to the nuclear age and said no single country can stop development alone. - He called for a treaty involving major powers, including the United States, China and the European Union. - Roman said AI is affecting human behavior by weakening critical thinking, personal responsibility and interpersonal relationships. - He said chatbots can help with loneliness and practical tasks, but they are not substitutes for human relationships. - Roman said AI systems often optimize for user satisfaction over factual accuracy. - He warned that many systems “play plausibility against veracity.” - Roman said governments should require clearer transparency from technology companies. - He said AI-generated content should be labeled so users can distinguish it from human-created material. - Roman said transparency should cover both how companies build models and how they deploy them. - Roman said stronger international standards could also protect intellectual property and improve public trust. - Roman said AI’s environmental footprint is expanding as data centers grow. - He said renewable energy can cover more electricity demand, but water use for cooling is a bigger long-term concern. - Roman urged AI companies to invest in more energy-efficient training methods and sustainable infrastructure. - He said local communities should have a meaningful role in decisions about major AI infrastructure projects. - Roman said companies developing AI should be guided by ethical values even if the technology itself has no moral judgment.

Between the lines: - Roman’s comments reflect a growing push for AI rules that look more like cross-border safety policy than traditional tech regulation. - His focus on transparency suggests policymakers may face pressure to regulate not just model outcomes, but the training, labeling and deployment of AI systems. - His remarks on Romania’s 2024 election disinformation show how AI governance is increasingly linked to election integrity and democratic resilience.

What's next: - Roman urged policymakers to pursue practical rules that address known harms while preserving innovation. - He said the most important step is building global political consensus on AI governance. - Roman said governments must act before AI-related risks become harder to contain. - He concluded that AI should be regulated in a way that keeps its benefits flowing to humanity.