Postcard from Paris: politics vs policy

BONJOUR. THIS IS DIGITAL POLITICS. I'm Mark Scott, and like many of you, I'm part of the X-odus to BlueSky. You can follow me here. I look forward to future episodes of this newsletter dissecting the inevitable fall of another social network.

— The French coalition government may fall as early as this week. That hasn't stopped Paris from flexing its muscles on digital policy — with varying results.

— Australia just banned anyone under 16 years old from having a social media account (sort of). This is a bad law that will lead to bad outcomes.

— Accusations abound that TikTok skewed the first round of Romania's presidential election for a far-right candidate. So far, there's no actual evidence to prove that.

Let's get started:

French digital swagger takes a hit

HELLO FROM THE EUROSTAR LOUNGE in Paris where I'm heading back to London after speaking at an event on how Europe can reduce (or not) its digital dependency on the United States. As we enter the week of Dec 2, the country's coalition government — led by archetypal French politician Michel Barnier — is about to be toppled over a political fight about its proposed budget. I'm no expert on French politics (for more on that, see here, here and here). But my time in Paris this week is a clear reminder, just like I outlined last week when it came to Berlin, that France is struggling between its aims for tech policy and the political reality in which French president Emmanuel Macron finds himself.

Much of this inevitably comes down to politics. Again, I'm not going to delve into the inner workings of why Marine Le Pen’s far-right National Rally political party has gained so much traction domestically (more on that here). But what we're seeing in France, like what we've seen in almost every country's election this year, is incumbents getting a kicking at the ballot box amid ongoing domestic anger against skyrocketing costs, a sense the Establishment has failed voters and a willingness for citizens to embrace populism.

But what does this have to do with digital policy? France has been the loudest advocate for a more insular approach to tech regulation and investment that focuses on championing local (read: French) champions to take on global competitors, primarily from the US and China. It's why Ursula von der Leyen, the returning president of the European Commission, embraced that concept in 2014 when she relied heavily on Macron to secure her position in the Berlaymont Building in Brussels. It's also why there were continued efforts, over the last five years, to tilt European public investments toward French priorities like the creation of European Union-based cloud computing infrastructure. Why? Because Macron had the political muscle to (mostly) get what he wanted — even if other EU countries opposed such a Gallic vision of digital policy.

Before I get angry emails, I understand that I'm over-simplifying the political dynamics at work here. But it's also true that, since Macron suffered a dual political humiliation during, first, the European Parliament elections in June and, second, in domestic elections a month later, he is not the same omnipresent shadow on tech policy as he once was. Again, this isn't just a digital issue. The French president's hobbled status — especially in light of the political meltdown engulfing Germany — also has knock-on effects on the EU's approach to Ukraine, climate change and immigration. But more than one EU policymaker has told me, in recent weeks, that a weakened Paris provides greater breathing space for Brussels to take a somewhat different approach to critical digital policy issues than would have been possible when Macron was in his full pomp.

What does that look like? In Paris, officials are abuzz about the upcoming AI Action Summit, or spectacle on Feb 10-11, 2025 that is the third global AI gathering after previous editions in the United Kingdom and South Korea, respectively. Interestingly, the French are shifting away from a sole focus on safety — much to the annoyance of London where officials still believe these summits are their brainchild.

Instead, the focus will be on artificial intelligence's impact on 1) work; 2) innovation and culture; 3) the 'public interest;' 4) trust; and 5) governance. Expect some high-profile sums of money to be announced around these topics — as well as glitzy presentations across Paris to showcase France as, legitimately, the center of AI for Europe. In reality, Macron and his officials are using the now annual AI gathering as a showcase for France, Inc. The hope: to drum up significant amounts of inward investment and secure the country's position as one of the global leaders on everything from generative AI (hello, Mistral AI) to AI governance (the Organization for Economic Cooperation and Development, whose work on that topic has been world-leading, is based in the French capital.)

This all sounds like top-level Macron peacocking. All world leaders want to be viewed as championing the most interesting causes. And, in AI, the French president can be associated with something that has caught the public's imagination ever since OpenAI released ChatGPT just over two years ago. But what will likely go unsaid in February — especially if the coalition government led by Barnier, the prime minister, falls this week — is that Macron's political capital is not what it once was. It's one thing to hail France as an AI champion when everything is going your way. It's something else when you invite the world's great-and-the-good (and me, hopefully) to Paris amid swirling questions about whether you're a lame duck president.

This divide between politics and policy also can be felt elsewhere. The French have been vocal champions, in Brussels, for the so-called Digital Networks Act. Those are efforts to rewrite how the Continent's telecommunications infrastructure is paid for to force US tech companies to fork out extra cash to support the bloc's legacy carriers. It's a political hot potato that has pitted the likes of Telefónica, the Spanish carrier, and Deutsche Telekom, its German rival, against some of Silicon Valley's biggest names. Yet without a strong Paris to nudge other European capitals to support these proposals, Brussels is shifting away from a hard-nosed approach that, in essence, would require the likes of Meta and Alphabet to hand over some of their profits to cash-strapped European telecom giants.

The list continues. France's incoming European commissioner Stéphane Séjourné — in charge of 'prosperity and industrial strategy' in Brussels — wants to promote a "Europe first" industrial strategy, in part as a response to Donald Trump's upcoming second term in the White House. To be fair, the former French Europe minister acknowledges the bloc must remain open to outsiders. But Séjourné's position would be stronger, in likely tense negotiations in Brussels on how to define such a 'Europe first' strategy, if Macron's own standing in Paris was more secure than it currently is.

How will this play out in the months ahead? A lot will depend on whether Barnier's coalition government survives (my bet is that it won't.) France's next presidential election will be held in 2027, meaning Macron will stick around — albeit in a more muted position than he once had. That will likely tilt the scales, at the least on tech policy, away from a more Paris-led interventionist/'let's pick winners' approach from Brussels where political leaders are already shifting away from digital policy to focus on defense and climate change.

The one major unknown, though, is what happens in Germany after that country's elections in February. If the EU's two most important countries fall into political chaos, all bets are off. If that does happen, expect the European Commission to strike out on its own to define the next five years on tech policy — without waiting around to see what unfolds in either Paris or Berlin.

Chart of the Week

When Călin Georgescu, a pro-Russia ultranationalist, went from nowhere to winning the first round of Romania's presidential election at the end of last month, many were in shock.

Attention quickly turned to how Georgescu used social media — particularly TikTok — to reach would-be voters and how his online presence may have been promoted by either domestic or foreign actors.

One of the country's regulators even asked the European Commission to look into it via the bloc's Digital Services Act.

But despite the smoke, there's currently no fire to suggest Georgescu had significant outside help — as you can see from the analysis, below, of all the main TikTok accounts from Romania's leading presidential candidates.

Since these figures were produced (just before the Nov 24 vote), Georgescu's TikTok following has jumped to 519,000. But his opponents also had sizable audiences on the China-linked platform.

"ExoScore" refers to a combined metric, based on how active the account is, the size of the account's following, and how much engagement the account receives.

Why banning social media for kids is counterproductive

WHEN I MET MICHELLE ROWLAND, the Australian politician in charge of the country's online safety regime, earlier this year, the former telecom lawyer didn't beat around the bush. "More needs to be done by platforms to keep Australians safe," she told me. Fast forward to late 2024, and Canberra just passed new rules to ban anyone under 16 years old from having a social media account. Here's the amendment to the country's Online Safety Act. It's the first time, anywhere in a Western country, that politicians have gone so far to stop minors from accessing social media. For campaigners, the updates — which carry potential fines of tens of millions of dollars if tech companies fail to act — are long over due to keep children safe online.

Unfortunately, that approach is wrong. We'll get into the inner workings of the law in a minute. But the main problem with Australia's new law is that it fails to tackle the underlying issue with social media, while treating all children with a level of paternalism that is out of touch with how kids use technology in 2024. Instead of forcing tech giants to take a 'safety-by-design' approach to how they develop so-called recommender system algorithms and thwart online hate, Australian lawmakers decided to create a binary choice that will have little real impact on how those under 16 years old access and consume social media content.

When it comes to kids' use of the internet, things get difficult very quickly. Almost all tech companies require users to be at least 13 years old to have an account, though how firms verify those ages leaves much to be desired. Parents should make the right choice for their own families on how children both access devices and online services. Most, let's be clear, are clueless on both fronts — often because they themselves are addicted to smartphones/social media accounts in a way that bleeds into how their kids similarly want access. High-profile and tragic incidents like the death of British teenager Molly Russell have led to a global movement, by some parents, to ban kids' access to both phones and social media.

I sympathize with many of these points. But Australia's new laws highlight how people's goodwill to protect children can often get lost in the granularity of how such blanket bans actually work. When Canberra's ban comes into force (most likely within 12 months), it will be down to the social media companies to determine which of their users are under the 16-year-old threshold. How will they do that? Either by requiring everyone to provide government-issued identification; relying on people to upload a credit card as proof of age; or trialing untested facial recognition technology.

For why all of those options are bad, read this. But my fear is that, in the knee-jerk effort to keep children safe online, Australia is removing a level of anonymity from online users (via age verification requirements) that can quickly lead to massive potential government overreach of the digital world. Worse still, dubious facial recognition technology — if rolled out poorly — could lead to significant harm if such sensitive and personal data is not protected securely, at all costs.

The questions just keep coming. Australia still hasn't determined which social media companies should comply with the age standards — and apparently YouTube may may be out of scope. Cue: endless lobbying as tech companies do whatever they can to define their services so they don't have to comply. The law also doesn't detail if messaging services like WhatsApp or gaming platforms like Microsoft's Minecraft will be in scope. The Online Safety Act amendment is so wooly on detail that much will be left to local regulators/officials to figure out on the fly. For me, that's not how good policy should be developed.

What's a better option? Let's take, as our starting point, that social media comes with both harms and benefits. As much as children may be shown graphic images, via their feeds, they also may find supportive online communities at a time when their offline worlds are imploding. To suggest it's either one or the other is to negate how social media actually works. A second point: we should protect children, both online and offline. But reducing harm for all using these services is also something to be championed.

That's why creating a basic principle, for all tech companies, that they should embed safety and privacy in all their services — and that they will be policed to ensure compliance — would be a more effective way to reduce social media harm.

What does that look like? Demanding that social media algorithms aren't tweaked to maximize engagement (often by displaying polarizing or harmful content) would be a start. So, too, would developing policies that require companies to reduce people's addiction to social media (by dampening down the dopamine-style user interfaces, for instance.)

That's not as binary as a 'let's ban social media!' policy that is easily digestible to voters whose fears for how kids use devices and social media are growing. But it would lead to what those concerns are based upon: a growing realization that something must change in how we all interact with digital services that now dominate our daily lives.

What I'm reading

— Stanford University's Institute for Human-Centered Artificial Intelligence published its annual AI Index Report. There's a lot to unpack. Enjoy.

— The European Commission asked TikTok to provide more information on how it handled potential manipulation of its platform. More here.

— Interpol announced its partners in 40 countries had arrested more than 5,500 individuals and seized more than $400 million as part of a major global cybercrime operation. More here.

— The US Commerce Department published an overview of last month's gathering of domestic AI Safety Institutes in San Francisco. More here.

— Companies published their inaugural risk assessments and audits under the EU's Digital Services Act. I've put together a running tally of all the documents here.