Description

As chatbots and conversational AI interfaces evolve into autonomous, agentic systems, the question is no longer only whether we can trust what machines say, but whether we can trust what they do on our behalf. In a geo-political context where adversarial actors actively seek to distort data and narratives – targeting both citizens and machines – the foundations of public trust are challenged and public-sector AI faces new responsibilities. This keynote speech explores how these changes are reshaping the meaning of trust, why authenticity and provenance must be built into conversational systems and generative AI by design, and what this shift means for public institutions. The Publications Office of the European Union has a critical role to play: anchoring AI-enabled services in legally authentic, multilingual, and trustworthy knowledge, while calling on technologists, industry, and policymakers to share responsibility for building AI that strengthens, not erodes, the rule of law, democracy and respect for human rights.