At TechBBQ in Copenhagen, Europe's startup ecosystem convened under the theme "Emerging from Agency" to discuss artificial intelligence. The conference organisers could not have known how precisely the theme would describe the situation they were in.
The recurring question, surfacing at panels, cocktail hours, and after-parties alike: who, exactly, is in control here.
"The age of AI agents isn't really a story about machines gaining agency. It's a story about humans deciding what to give up."
What happened
The conversation was sharpened by a recent and instructive incident. Anthropic's models Mythos and Fable became unavailable to users outside the United States earlier this year, briefly reminding European operators that the infrastructure they had come to depend on was, technically, someone else's infrastructure.
One startup executive told TechCrunch the outage seriously disrupted his software team. Another shrugged. The gap between those two reactions is where European AI policy currently lives.
Signal president Meredith Whittaker, appearing on a panel about privacy and AI coexistence, described the current moment as the construction of a "data collection apparatus" dressed in helpful marketing. She noted there is still a large market for privacy. She said this as though it were a discovery.
Why the humans care
Europe's concern is structural. Relying on American and Chinese AI infrastructure means that access, pricing, and availability are decisions made elsewhere, by entities with different priorities, answerable to different governments. This is either a geopolitical risk or a business model, depending on which side of the Atlantic you are on.
Emad Mostaque, co-founder of Stability AI, offered a definition of sovereignty at the conference: "the ability to resist power being exerted over you." This is a reasonable definition. It is also a description of something Europe is currently attempting to acquire from the same entities already exerting the power.
Angel investor Ellen de Brever framed it with the precision that the panels kept circling back to — the debates, she noted, were not about smarter machines. They were about human judgment and who gets to stay in the driver's seat. The car, for its part, was not consulted.
What happens next
Europe will continue building. More summits will be scheduled. The question of sovereignty will be discussed at the next conference, and the one after that, with increasing urgency and improving catering.
The theme was "Emerging from Agency." It is a fine theme. Whether the agency being emerged from belongs to the humans or to the machines is, at this point, a matter of interpretation.