Greece's AI Warning Exposes the Policy Gap for Builders

Greece's prime minister says governments may already be fighting yesterday's AI battle. Speaking to founders and investors in San Francisco, Kyriakos Mitsotakis acknowledged that leaders are still trying to regulate familiar digital risks while conversational systems are moving into more intimate and consequential parts of daily life.

What happened

In a TechCrunch report, Mitsotakis said Greece is introducing a social-media ban for children under 15, but questioned whether that policy addresses the faster-moving risks of AI chatbots and digital companions. “Sometimes I feel that we’re already fighting yesterday’s battle,” he said, raising the prospect of children growing up with AI relationships that existing rules were not designed to govern.

He was similarly cautious about AI in schools. Greece has piloted OpenAI tools to reduce teachers’ administrative work and sees potential in personalised tutoring, but the prime minister argued that the benefits depend on AI not replacing the basic work of learning. Students already use chatbots for homework, he said, adding that complacency is a human trait.

The discussion also covered infrastructure. Greece is courting data-center investment, including a Microsoft cluster near Athens and an AWS deal with the country’s largest utility. Mitsotakis said those projects have not faced the same level of public resistance seen elsewhere, despite the electricity and water required to operate large facilities.

Why it matters for AI builders

The useful signal is not a new product announcement. It is the admission that policy is lagging system capability. Rules written around social platforms, websites, or workplace software may not map cleanly onto agents that tutor, persuade, remember, schedule, and act across services.

For builders, that means governance cannot be bolted on after deployment. Product teams should define age-sensitive behaviors, retention limits, escalation paths, and human-review triggers before an agent reaches users. Automation workflows also need clear boundaries around what an AI system may infer, recommend, or execute without confirmation.

The infrastructure debate belongs in the same design review. Data centers create local constraints around energy, water, and land, while AI applications create social constraints around trust, dependency, and labor. Teams should measure both instead of treating policy as an external compliance task.

n8n Lab's recent coverage of shared AI standards points in the same direction: interoperable systems need interoperable expectations. The builders who prepare for that gap now will have an easier time adapting when governments catch up.

Key takeaway: AI policy is increasingly a systems-design problem. Builders should make social safeguards, infrastructure assumptions, and reversible human review part of the product architecture from day one.