{"text":[[{"start":4.56,"text":"The writer is a fellow at Stanford University’s Institute for Human-Centered Artificial Intelligence and author of ‘The Tech Coup’"}],[{"start":16.88,"text":"European Council president António Costa is scheduling a discussion among European heads of state about AI this autumn. Better late than never. Leaders must use this opportunity to catch up and align on AI policy. While the US and China treat the technology as a key priority and instrument of state power, the EU has not even decided on a collective political course."}],[{"start":39.24,"text":"Today, any discussion of digital sovereignty or strategic autonomy is, in effect, a discussion about AI. This global gamble creates a risk for Europe. Without sovereign AI options, Europeans are dependent on foreign platforms and tools. These dependencies go beyond commercial decisions. They mean that decisions made by foreign governments have a disproportionate impact on European national security."}],[{"start":64.84,"text":"Political leaders must therefore confront the AI side of security head-on as tension builds between AI labs and geopolitics. They should prioritise trust and safety."}],[{"start":75.56,"text":"In the US, where investment in AI is breaking all records, the Trump administration regards AI as a source of its hegemonic global ambitions. President Donald Trump is taking ever more steps to boost US advantages in AI capacity, access and agency. Reports suggested that Anthropic’s latest Mythos AI model was being used by US intelligence agencies, while access for European allies was limited."}],[{"start":100.26,"text":"This presents a new security challenge for the EU. The recent Nato summit in Ankara saw European leaders confront the reality of an unreliable and possibly adversarial America that forces the EU to develop and maintain its own military and defence capacities. Cybersecurity is integral to resilience. Technological developments also make it clear that divorcing economic choices from security objectives is impossible. Corporate actions have an increasing impact on safety and security."}],[{"start":128.68,"text":"AI is at the heart of this inconvenient truth. To tackle potential threats from cyber incidents or exploitations of dependencies, EU states must set up structures to co-ordinate emergency responses and benefit from sharing lessons learnt. The summit should decide on those."}],[{"start":145.12,"text":"Dedicated task forces have created urgency and focus during past crises in Europe, including Covid and the financial crisis. They are merited here too. A clear AI policy package should include hiring AI talent into policy positions and developing national AI security institutes, as well as a broader talent programme to grow and attract AI talent (back) to Europe."}],[{"start":166.52,"text":"Within the continent there already exists tremendous energy and appetite for stronger EU leadership on AI. National AI strategies have been announced and adopted, some universities are decoupling from non-EU tech providers and intelligence agencies are choosing domestic systems over those offered by international companies. Tying such bottom-up decisions with an EU-wide vision should scale and boost the bloc’s competitive position by dividing labour between member states and leveraging capital incentives, ideally in sync with industry. European leaders can do a real service to people worldwide who are deeply unhappy with the two-flavour menu of AI they are currently offered."}],[{"start":206.52,"text":"Being ambitious politically does not mean prioritising AI above everything else. Yes, Europe can choose to do AI differently and not copy the extractive Silicon Valley way of reducing people to consumers and societies to markets. Europe is and will remain far removed from the Chinese model of state-directed companies and innovation. But being principled and right on paper translates into very few results in today’s data, capital and compute-driven world. If Europeans believe AI can add value while respecting democratic values and increasing safety, they must show so."}],[{"start":242.56,"text":"The summit this autumn must have a clear agenda and real ambition from which European leaders walk away with a comprehensive AI strategy."}],[{"start":250.38,"text":"It should therefore conclude with serious political buy-in and a tangible strategy that unites the leverage that comes from Europe’s single market and wealth: incentivising investment into sovereign capacity such as compute, making AI chokepoints more resilient, and setting priorities for which capacities must be nationally held first."}],[{"start":269.44,"text":"Any strategy is only as good as its execution. The summit must provide conclusions and actions. The EU cannot afford to waste this opportunity and spend the autumn merely learning and brainstorming about AI."}],[{"start":285.46,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1785986514_5849.mp3"}