11 min read

What you need to know about France's AI Action Summit

The upcoming summit in Paris will include efforts to boost AI governance. But for Emmanuel Macron, it's all about promoting French industry.
What you need to know about France's AI Action Summit
This image was created via DALL·E

BONJOUR, MES AMIS! I'm Mark Scott, and will be heading to Paris on Feb 10-11 for the upcoming AI Action Summit (more on that below.) If you're also going and want to grab a coffee (or croissant?), reach out here.

Also, for people in Washington, I'm teaming up with Katie Harbath (and her excellent Anchor Change newsletter) for a tech policy event in Washington the week of March 10. If you're interested, let me know here.

— The French pow-wow on artificial intelligence should be seen for what it is: an effort by the country to position itself a global AI leader.

— Here's a new concept you're going to hear a lot about in the years ahead: "Euro stack." Let's unpack what that actually means.

— Just under 20 percent of teenagers are now addicted to YouTube and TikTok. Don't believe me? Check out the chart below.

Let's get started:


We're good at AI too, say the French

ON FEB 10-11, EMMANUEL MACRON, the embattled French president, will host heads-of-state, policymakers, tech executives and civil society groups (and me) at the AI Action Summit in Paris. It's the third iteration of this now-regular summit that the United Kingdom kicked off, in late 2023, and then the South Koreans continued last year. Expect a Global Majority country (my bet is on India, a co-host for next week's conference in France) to host the subsequent event, most likely in early 2026.

First, the basics. On day one, there will be a series of official events (full agenda here) on everything from international AI governance to the emerging technology's impact on the workforce to its environment footprint. Expect a lot of talk about "inclusivity," "innovation," and "trustworthy AI." "We must enable artificial intelligence to fulfil its initial promise of progress and empowerment in a context of shared trust that contains the risks inherent to technological development, while seizing every opportunity," according to the French government. Cue: AI policy buzzword bingo.

Day two is just for governments. The rest of us will scatter across Paris for side events on topics like AI's impact on the information environment, trust and geopolitical relations. Countries will then publish a summit communiqué — akin to previous summits (here and here.) I wouldn't expect much. Based on the French government's public statements — officials have been traveling the world ahead of the February event — I would expect a reaffirmation of embedding human rights and openness into AI's development; the need to promote innovation without allowing a few (American) firms to dominate; and tackling the environmental and social impact of a technology that has caught the public's imagination.

As the upcoming event will be held in Europe, I would also put good money on at least a name-check to greater AI governance and regulation. "The Summit will therefore reflect a balanced European approach to artificial intelligence that combines support for innovation, adequate regulation and respect for rights," based on France's stated objectives. That's somewhat ironic after Macron tried to water down the European Union's Artificial Intelligence Act, at the last minute, over fears it would hobble the country's nascent AI industry. Those comprehensive rules won't come into full force until late 2026. So, for now, Paris is willing to at least publicly support legislation that, privately, it remains skeptical about.

Thanks for reading the free version of Digital Politics. Paid subscribers receive at least one newsletter a week. If that sounds like your jam, please sign up here.

Here's what I wrote about in January:
— The lessons platforms learned about Jan 6 riots was to pull back on content moderation; 2025 will see a stalling of AI governance. More here.
— The proposed TikTok ban isn't about free speech or national security. It's about the geopolitical clash between Washington and China. More here.
— How Brussels will respond on digital regulation to the Trump 2.0 administration; Community Notes aren't good at fact-checking. More here.
— A Who's Who guide to tech policy officials in Washington; Why national security doesn't make for good digital policymaking. More here.
— The United Kingdom does not have a clear strategy when it comes to digital. More here.
— In the global AI fight, bigger (infrastructure) doesn't always mean better results; Why transatlantic data flows are again in jeopardy. More here.

What won't be a priority for France is AI safety. That was the sole focus when the UK started this summit-a-palooza 18 months ago. During that event just outside of London, the then-British prime minister Rishi Sunak went hard on the existential risk of the technology, including the creation of the country's AI Safety Institute. In Seoul last year, the South Koreans (with the somewhat strong-willed support of the Brits) shifted to include "inclusivity" and "innovation." On Feb 10-11, the French will go hard on that last concept, relegating AI safety to an also-ran concept lumped into wider discussions around governance.

It's not that Macron & Co aren't concerned about AI's downsides. But they haven't fallen completely — as the former UK government did — for the belief that the existential threat of the emerging technology is the main long-term risk. For Paris, the consolidation of power, including within the underlying infrastructure required to build next-generation AI systems, is a more paramount threat. That's why you'll hear a lot next week about so-called "public interest AI," or a more inclusive, decentralized version of how AI can developed. One that is based on open source technologies, a community-led approach to solving societal problems and a counterweight to the Silicon Valley tech bro brigade.

What's not to like, right? Well maybe. The Feb 10-11 summit also provides France, Inc with an opportunity to flog its wares to a global audience descending on Paris to talk AI. And there's a lot to flog. The French AI tech darling Mistral will get more shout-outs than people saying "autre vin rouge, s'il vous plaît." (That's my last French stereotype, promise.) But the country has world-leading research hubs in places like Lyon and Toulouse. Both Alphabet and Meta have separate AI research teams in Paris. In an event — entitled "Business Day" — on Feb 11 at Station F, a sprawling startup center in the French capital, local techies will vie for attention as part of the country's wider efforts to pitch itself as the center of Europe's AI industry.

Again, there's nothing wrong with some American-style bravado to celebrate France's local AI champions. But it's not exactly what these summits were supposed to be. The UK may have gone too hard with its AI safety focus in 2023. But it was at least an effort to bring countries, including China, into a room to talk through how to collectively combat the doomsday scenarios. Fast forward 18 months, and the AI Action Summit is now more a roadshow for Macron to drum up foreign direct investment. Concepts like governance, inequalities and sustainability — ideas that are, in principle, still part of the event — have been quickly overshadowed by the unending need to boost France's domestic economy.

Before I get angry emails from French officials, the Summit's communiqué, based on public statements from the country's officials in the build-up to the event, will likely still highlight the wider societal goals of AI governance. I would be particularly focused on what may come from any efforts to promote public interest AI as a counterweight to the growing concentration of economic power among a few Silicon Valley giants (and China's Deepseek, if you believe the recent hysteria.)

In 2025, my bet is that further government oversight of the emerging technology will be put on the back burner in the name of global competitiveness. That will even happen in places like the EU and South Korea where lawmakers have passed comprehensive AI rules.

In that context, a voluntary statement from countries, in the form of an AI Action Summit communiqué, won't be worth much.

A more positive view of next week's summit is that France is actually building AI products, based on governance principles, instead of just talking about the need for oversight. A more negative perspective is that the Feb 10-11 conference is an effort by Macron — suffering from shifting political winds at home — to regain the advantage by demonstrating his role as a global leader on AI.


Chart of the week

POLICYMAKERS WORLDWIDE NOW OPENLY fret about how addicted children have become to digital services. Some, like those in Australia, have gone so far as to ban access to social media platforms for minors.

To be clear, there is no empirical evidence to connect growing levels of mental health illnesses, among kids, to access to digital services. That doesn't mean children should be let loose on the likes of YouTube and TikTok.

And yet, roughly 15 percent of American teenagers are now almost constantly glued to those services — with addiction to Instagram and Snapchat (but not Facebook) not far behind.

Source: Pew Research Center

Make Europe Great Again!

THERE'S NO DENYING GEOPOLITICS has taken over technology. The United States now vies openly with China on everything from high-end semiconductors to critical raw materials. In that bipolar world, Europe — and its focus on principles-based digital regulation that promotes fundamental rights — may represent a third way, according to Anu Bradford, who coined the "Brussels Effect" concept.

Yet there is now a rival theory about how the EU can compete globally that's gaining traction in European policymaking circles. And that involves building a so-called "Euro Stack" of digital infrastructure, tooling and services that is both made within, and run solely from, the 27-country bloc.

For those who attended the inaugural "Marked As Urgent" tech policy event on Jan 30, thank you. You can see photos from the meet-up in London here. We'll have another event for you on March 27. Sign up here for details.

If you're interested in sponsoring or partnering with Digital Politics as I develop the newsletter and future events in 2025, please reach out at digitalpolitics@protonmail.com

Leading that charge is an Italian economist called Cristina Caffarra. She earned her spurs in the cut-and-thrust world of digital competition, advising companies like Apple, Microsoft and Amazon, as well as a series of European and US antitrust officials. Caffarra and others want an industrial policy to meet the new geopolitics where competitiveness and economic growth — as outlined in Mario Draghi's report for the European Commission last year — is the new name of the game. Donald Trump's return to the White House, from this worldview, has made the Euro stack more important than ever.

"It's a massive disgrace that when I have a video conference with the Commission, I use (Microsoft) Teams," Caffarra told an audience in Brussels on Jan 31. "Buy European. Europe First." That last comment received a massive applause from the European crowd. "The reality today is that we are a colony," the Italian economist continued. "The energy was focused on digital regulation as the only thing we had, it was a massive mistake." To make that point even clearer, the Italian also held her own conference — dubbed "The Perfect Storm: A Time of Truth for Europe?" — in Brussels on Jan 30.

There's a lot to unpack here. For those promoting the Euro stack concept, they worry about the dominance of American tech giants in key digital infrastructure areas like cloud and quantum computing. Without homegrown alternatives, the theory goes, Europe (and other parts of the world, too) will always be beholden to the US' commercial and/or political whims. To fix that, the EU must build its own rival infrastructure — preferably based on open source principles to avoid future industrial capture — to meet European needs. For what that could look like, see here and here.

I have some sympathy for that argument. But only to a point. Yes, there needs to be greater offerings from diverse actors when it comes to building the underlying infrastructure for the global digital economy. A reliance on a small number of companies — be they American or not — is not sustainable.

But where I disagree with the Euro stack pitch is its jingoistic approach that tries to Make Europe Great Again. We've already seen the bloc try to create its own version of Google. That failed miserably. Euro stack supporters would say this is about creating homegrown infrastructure, and not just replicating what already exists. Sure, I get that. But when I hear the likes of Caffarra speak, it sounds a lot like people complaining that Europe didn't get the economic bump from existing digital services. If Meta was, for instance, based in Paris and not Menlo Park, would they have a similar critique of the dominance of the online world by a small number of — in this alternate reality — European champions? I doubt it.

There's also a misreading of the Euro stack crowd of what India achieved with its own version of this concept. For more on the so-called "India Stack," read this and this. But, in essence, New Delhi created a series of easily-accessible public data access points on which private companies and the government could then provide new services. That has led to problems, most notably around people's privacy. But — and I'm not an expert in this policy area — India's approach to create homegrown alternatives was more about opening up existing data, which had been siloed, for new commercial and social opportunities. It was not, as envisioned in Europe, as a like-for-like retrofit of existing (mostly American) infrastructure for domestic alternatives.

I have more questions. If the Euro stack is about investing billions, if not trillions, of dollars in European-owned infrastructure, who is going to pay for it? And if such alternatives can be funded — most likely via public resources, given that buckets of private capital have already had years to invest in this opportunity, but didn't — are we OK that citizens will likely pay more compared to what they already have access to via existing infrastructure? Even in the current more transactional geopolitical environment, is Europe willing to put up the borders to outsiders — even if they can offer European citizens (cheaper) services that meet their needs?

My largest criticism of the Euro stack movement is not their frustration with the status-quo. I get it. American tech companies now dominate much of the digital world (outside of China.) To boost Europe's long-term economic and societal interests, reducing that dependence makes good politics.

But in their breathless attempt to frame the existing situation as a mere failure of digital regulation and an unwillingness of EU officials to get tough against the US, the likes of Caffarra are missing how you "win" (note: I wouldn't view this as a zero-sum game) in the global fight over digital.

If their underlying criticism is of an industrial model that has reinforced power around a small group of Silicon Valley giants, you don't overcome that by replicating such structures — but just with French, German or Swedish firms. You do it by taking what non-Europeans (including non-US countries like Japan and South Korea) do best, and overlaying that with local solutions created, and championed, by local citizens.


What I'm reading

— Chinese covert influence operations impersonated human rights organizations critical of Beijing to discredit these groups' activities, according to a report from Graphika.

— The law firm Arnold & Partner analyzed what Trump's new executive orders on AI mean for developers of this emerging technology. More here.

— The European Commission announced a series of measures, called the Competitiveness Compass, to boost the bloc's growth. More here.

— Elon Musk remains immensely unpopular in both Germany and the UK despite his efforts to wade into those countries' domestic politics, according to a YouGov poll.

— The Chinese large language model DeepSeek performed well when a research asked it to respond to X posts as if it was a propagandist for the Russian government. More here.