Superintelligence Europe — No. 005

Teresa Ribera, the European Commission’s Executive Vice-President for competition, travelled to San Francisco on Tuesday for the most significant face-to-face engagement yet between Brussels and the architects of American AI. Her schedule included first-time meetings with Alphabet CEO Sundar Pichai and Meta CEO Mark Zuckerberg, as well as OpenAI CEO Sam Altman — all in the same week, on their home turf, as Ribera carries active investigations into two of the three companies. Amazon CEO Andy Jassy is scheduled for Wednesday.

The meetings are not courtesy calls. Ribera has said she is examining the entire AI stack — chatbots, training data, cloud infrastructure, and the energy sources beneath it. The Commission has a Statement of Objections against Meta for its WhatsApp AI restrictions, is investigating Google’s advertising technology under the Digital Markets Act, and has signalled that a DMA ruling on Google is imminent. Ribera told the Wall Street Journal before departing: “It will come” — a phrase that landed heavily in every European legal and tech affairs desk. Her trip also follows a direct confrontation with the US over regulatory sovereignty, after she described US pressure to soften EU digital rules as “blackmail.”

A new AWS-backed study published Tuesday found that 54 percent of European businesses now use AI — up from 33 percent just two years ago. That headline number tells an optimistic story. The paragraph beneath it does not. Advanced AI usage, defined as embedding AI into core processes, building custom solutions, or deploying systems capable of planning and executing complex workflows, has barely moved. It increased just one percentage point over the past year to reach 22 percent. The gap between basic and advanced adopters is not narrowing — it is widening.

The productivity implication is stark. Advanced adopters are 55 percent more likely to report significant productivity gains than those stuck at basic use. If Europe’s 58 percent of basic adopters could be moved to advanced use, the study estimates €191 billion in additional gross value added would be unlocked for the European economy. The agentic AI data point is the most striking: only 24 percent of European businesses have even heard of agentic AI, and just 3 percent of those familiar with it have deployed it.

OpenAI announced on Tuesday that it is shutting down Sora, its AI video generation platform, six months after launching a standalone app for the service. The company said it is refocusing resources on coding and core business priorities ahead of a planned IPO. The closure immediately triggered the collapse of a $1 billion investment and licensing agreement with Disney, which had been expected to bring the company’s intellectual property to the Sora platform. Disney confirmed its exit, framing the decision around protecting creator rights and intellectual property.

OpenAI is not exiting AI video entirely. The capabilities are being folded into ChatGPT rather than maintained as a standalone product. But the strategic signal is clear: the company is narrowing its product surface ahead of its public offering, and video generation — resource-intensive, legally contentious, and commercially unproven at scale — did not make the cut.

Euronews devoted Tuesday’s flagship interview series to a question that European health ministries have been quietly asking for years: whether AI can absorb enough of the structural pressure on European healthcare systems to prevent the kind of capacity collapse that demographic ageing and chronic underfunding are making increasingly plausible. The CEO of HealthAI argued that the current moment is a genuine inflection point — that AI offers the first realistic pathway to lowering the disease burden at population scale while reducing per-patient cost.

The case rests on a specific vision of AI integration: not chatbots as triage tools, but AI embedded in diagnostics, clinical workflow, drug development timelines, and hospital operations — compressing the time between symptom and treatment, and between research and deployment. The practical barriers in Europe are significant: fragmented health data across member states, strict GDPR constraints on patient data sharing, legacy hospital IT infrastructure, and procurement processes that move slowly relative to the pace of AI development. The conversation on Tuesday was aspirational. The implementation challenges are not.

Credible accounts and publications that drove Tuesday’s European AI conversation. Filtered for genuine signal.