POLITICO’s weekly transatlantic tech newsletter for global technology elites and political influencers.

By MARK SCOTT

Send tips here | Subscribe for free | View in your browser

THIS IS DIGITAL BRIDGE. I’m Mark Scott, POLITICO’s chief technology correspondent, struggling to file this newsletter because of a bad Wi-Fi connection on the Eurostar heading back to London from Brussels. Meanwhile, here’s another example of AI taking over the world.

It’s a good one this week. Promise.

— People’s hyperventilating over artificial intelligence undermining elections worldwide needs to stop. Yes, the threat is real. But the impact is very much unknown.

— Margrethe Vestager, Europe’s digital chief, spells out what she wants her successors to focus on over the next five years, and what her kids think about her smartphone obsession.

— India heads to the polls this weekend. The outcome isn’t in doubt, but the digital campaigning tactics are complex, opaque and increasingly sophisticated.

BOTS AND BALLOTS: WHAT YOU NEED TO KNOW ABOUT AI AND ELECTIONS

I DON’T KNOW MANY THINGS. But one area where I have some expertise is disinformation’s impact on elections. And in this year’s motherlode of global elections — happening in over 50 countries, encompassing more than 2 billion people — everyone appears to be freaking out about AI’s influence on the increasingly polarized voting populace. The theory goes that the technology’s rampant advances — coupled with the ability of bad actors to use it to peddle synthetic falsehoods across mostly unregulated social media — make 2024 ripe for AI-powered skullduggery. Look at the Joe Biden fake AI robocalls in New Hampshire; the spoofed AI audio clips in Slovakia and the United Kingdom; and the way Pakistan’s Imran Khan used so-called generative AI to run for office from jail. This stuff is real, and, potentially, harmful.

To that, I say: sure. If you look at the problem from a very micro level. But after talking to more than 30 politicians, policymakers, national security officials and disinformation experts — from many of the countries holding elections this year — the resounding question is: Where is the proof? It’s certainly true that more AI-generated content is making its way onto social media. Entities connected to both Russia and China now use it in their influence campaigns. Campaigns and their affiliates rely on so-called deepfakes (AI-generated images, audio clips and videos) to promote their interests. Governments don’t really know how to respond, technically, so have punted the short-term response to tech companies — many of which are the ones that created these AI tools in the first place.

But, again, I come back to the thorny question that no one wants to admit. For all the potential harm this technology may represent, no one has yet connected it to unfair voter outcomes that skewed an election in favor of one politician over another. As we’re almost four months into 2024 — and we’ve already had elections in, among other places, Bangladesh, Pakistan and Indonesia — it’s a legitimate question to ask about quantifiable evidence that AI is the harm that many say it is. I am a lapsed business reporter. So my adage remains: Politicians lie, numbers don’t. And, so far, no one has successfully demonstrated that AI has harmed a country’s electoral process in such a way that has broken underlying democratic institutions and processes. Might that happen? Totally. But as of April 18, 2024, we’re still in the “wait and see” stage.

Before I get irate emails, let’s be clear: No one may be able to prove such harm. But as politicians rush to legislate — or, at the very least, hold public hearings to chastise AI companies — we should be honest about what’s really going on. The emerging technology is definitely a new threat, in terms of allowing bad actors to create harmful, politically motivated content, at scale and at very low cost. But the experts I spoke with were very clear: It’s an additional tool in the existing arsenal of these players, and not a totally new thing that will shift political narratives from Berlin to Boston to Bangkok.

The obvious next question on my mind is: Why is that? For me, it comes down to four things. First, the technology is a work in progress. I tried to clone my voice, for instance, with mixed results (check that out here). Many of the existing deepfakes are either quickly debunked or exposed by technical errors, including efforts by companies to insert fingerprints into sythentically created content. That may change. But, for now, most AI content isn’t good enough to fool most people (try that out for yourself here). Second: You just don’t need AI to create and spread disinformation. The recent Gaza conflict — which has often resulted in gruesome posts flooding social media platforms — has seen its fair share of falsehoods. But most of these have been taken from other conflicts or video games. Why use AI when there’s existing nastiness that you can just repurpose?

Third is how voters think about elections. As much as I spend my time glued to social media (for work purposes, obviously), most people don’t. They also aren’t likely to change their entrenched political views because of a deepfake that they may have come across while scrolling. To suggest that voter habits can be duped by AI so easily underestimates the public’s political sophistication. The fourth reason for calming down about AI is more basic. Survey after survey has shown that the group of individuals most likely to peddle disinformation is — wait for it — elected politicians. Those lawmakers may seek to use AI to extend their footprint. But, for now, many have existing reach through large social media followings. That’s not what many politicians want to hear. But it’s where we are right now.

That doesn’t mean people are rightly afraid of AI and its impact on elections. My POLITICO colleagues and I brought together poll after poll that highlighted an aversion to the (often misunderstood) technology and its ability to power disinformation. But — bringing this back to quantifiable evidence over gut feeling — that doesn’t mean AI has unfairly shaped an election. Let’s be clear: That may happen in 2024. But amid the breathless hysteria surrounding the topic, it’s time to take a breath, look for evidence of harm, and stay vigilant.

MARGRETHE VESTAGER’S EXIT INTERVIEW

MARGRETHE VESTAGER IS AS CLOSE TO A MEGASTAR that Brussels policy circles can offer up. The Danish politician has been the European Commission’s competition chief for 10 years and, for the last five, the bloc’s top digital official. With less than six months to go before she steps down as Commission executive vice president, Vestager has a message for whoever comes next. “Of course, there will also be legislation in the next mandate, obviously,” she told me earlier this week, referring to the next Commission. “But my hope is just that they will focus on a lot of implementation because promises have been made, laws have been passed.” She talking about the litany of digital rules — the Digital Services Act, the Digital Markets Act, the Artificial Act, the Data Act, the Digital Governance Act — that Brussels has passed since 2019.

A lot of that comes down to money and personnel. Arguably, the European Union’s social media rules, digital antitrust overhaul and world-first comprehensive AI legislation are what most people will think about when it comes to this Commission’s digital legacy. Yet there are legitimate concerns Brussels hasn’t set aside enough people and cash to support those efforts that will likely be copied by other jurisdictions seeking a similar legislative path. “It’s always fair enough that a publicly-funded institution has resource limitations because that makes you make priorities,” Vestager responded — diplomatically — when I asked her about more resources for Europe’s digital rules. “But I do hope that resources will increase for these new enforcement tasks, especially because we get to things that are really close to the business model.” That last part refers to the bloc’s competition revamp that requires regulators to get ahead of potential abuses by checking companies’ expansion into as-yet uncharted markets.

Interviewing Vestager is always a poisoned chalice. Unlike most politicians, she is open, straight-talking and willing to listen to questions. But — not surprisingly for a politician who, until recently, had a mostly spotless reputation — she is a master of giving little away. One area where she and her global competition counterparts are focusing minds is how a cabal of mostly American tech companies are positioning themselves to become dominant in the emerging world of AI. Case in point: Microsoft’s “partnerships” with OpenAI and Mistral, as well as its so-called acquihire of the top talent at InflectionAI. “I think they are different. I think the Mistral deal is different from the OpenAI deal, which is different from the Inception deal. So they will work in different ways,” said Vestager. “And I’m just very careful to make sort of blanket judgments as to what will be the nature of the developing AI sector.”

OK, but what about competition concerns that these firms are sidestepping existing antitrust rules — still mostly focused on company A buying company B for a set amount — by partnering, not purchasing, smaller firms? “We [also need] a global approach because if entrenched market power needs something [like the EU’s Digital Markets Act] in order for the market to stay open, well of course, we may consider how can we even better use these tools if AI is entrenching market power even more.” Does that clear it up for you, in terms of what Vestager thinks about potential market concentration in AI? No, me neither. She can be slippery.

Inevitably, our conversation turned to geopolitics. The Danish politician has been central to the EU-U.S. Trade and Technology Council, met with her American counterparts in Washington last week, and gave a speech in Princeton that was hawkish on how to approach China. Vestager pushed back when I suggested that represented a shift by Brussels toward a more hostile stance favored by Washington. Europe, she responded, was doing its own work to decide which emerging technologies like semiconductors were too sensitive to share with the likes of China. “Otherwise, we’re very much a taker of a U.S. risk analysis as to what they find to be risky with their perspective,” Vestager said. “Sometimes (Europe’s analysis) will be different because this is Europe and not the United States.”

On China, Vestager reverted to her traditional talking point — despite the recent speech in the U.S. that portrayed the world’s second-largest economy as a strategic rival. “We should not ‘decouple,’ but we should ‘derisk,'” she said when asked about `Europe’s strategic positioning. “We do not talk a lot about the not ‘decouple’ (partly) because that is all the normal dealings with China that you do you know, clothes that produce in China, things are produced in China,” she added. “We talk a lot about what we can consider to be risky and, I think, that is why sometimes it can seem unbalanced.” That’s probably not what people in Washington want to hear. But, in Europe, there’s an entrenched divide between EU countries that have important commercial ties to China and those willing to be more aggressive with their dealings with China. As a representative of the EU’s nominal executive branch, Vestager has to walk that divide.

Given that this was likely my last chance to quiz Europe’s digital chief in person, I couldn’t help but ask about how her own use of tech has changed over the last decade. “I’m not as enthusiastic as I used to be,” conceded Vestager who as a politician in her homeland became one of the first to use Twitter (as it was then called) to communicate directly with voters. “We came into it with such an excitement, we were so encouraged. We found this was not new and shiny.” More importantly, she got some advice from her three daughters, all now in their 20s. “They just say, ‘Put down that phone, mom,’ because they don’t use it as much as I do.” That was good advice, she added. “The next generation, they say, ‘So what? It’s not safe. It’s stealing my time. I want to be with people, put down that phone.”

**How is artificial intelligence reshaping elections worldwide? POLITICO's Chief Tech Correspondent Mark Scott dives into the rise of misinformation, the impact of technology on democracy and more in our editorial series "Bots and Ballots", presented by Luminate. Read it here.**

BY THE NUMBERS

DIGITAL DEMOCRACY: INDIA EDITION

WHEN IT COMES TO ELECTIONS, they don’t come bigger than India. The world’s most populous country starts holding its nationwide vote on April 19 — with full results not expected until early June. Few expect anyone other than the current prime minister, Narendra Modi, to be the country’s leader after the dust settles.

It’s worth remembering that, despite this newsletter’s focus on digital, India’s election — and, frankly, most of the other votes this year — are mostly analog affairs. But a few tech-focused quirks caught my attention. The first was an 8-minute YouTube video from poet-turned-politician Muthuvel Karunanidhi that praised his son, another politician, ahead of this month’s vote. The thing is, Karunanidhi died in 2018 and the video was created via (bad) generative AI.

Modi’s Bharatiya Janata Party also has turned to AI to create an Instagram post of dead Indian singer Mahendra Kapoor singing the praises of the country’s incumbent prime minister. The video did not run with a disclaimer that it was AI-generated.

WONK OF THE WEEK

AFTER THE EUROPEAN DATA PROTECTION BOARD, a pan-EU group of privacy regulators, decided Meta couldn’t force people to choose between paying to use the company’s social media platforms or being shown targeted advertising, it’s time to focus on Isabelle Vereecken, head of that group’s secretariat.

Vereecken and her colleagues have been increasingly influential in determining how Europe’s data protection rules are implemented, often drafting pan-EU diktats for how the bloc’s data protection rules should work.

The Belgian lawyer has a long-standing track record. Before joining the secretariat, she spent more than a decade working for the Belgian privacy regulator and helped the European Commission to draft the EU’s General Data Protection Regulation (GDPR).

**Berlin Playbook, the newest addition to POLITICO’s Playbook family, launched! Täglich informieren wir Sie darüber, was am vor Ihnen liegenden Arbeitstag wirklich zählt. Die aktuellsten Ereignisse aus Kanzleramt, Bundestag und den politischen Zentren der Welt. Mit nur einem Klick anmelden.**

THEY SAID WHAT, NOW?

“We’re in the bottom of the second inning, and we’re ahead,” said Gina Raimondo, the U.S. commerce secretary, referring — using a baseball metaphor — to how much work the U.S. still had ahead of itself when doling out cash for domestic semiconductor production. “But there’s so much more ahead of us than there is behind us.”

WHAT I’M READING

— The ability for outsiders to independently verify Reddit’s data as part of wider transparency work is hampered by the company’s flawed design of these systems, which may lead to important information being missed, according to an analysis from the Citizens and Technology Lab.

— Stanford University published its annual Artificial Intelligence Index Report. It’s more than 500 pages. Take a look here.

— The United Kingdom’s online safety regulator published its research focus as part of its growing enforcement work, including understanding user behavior and what constitutes online harm.

— TikTok’s measurement of what appears on its popular social network differs markedly depending on whether you access that information via the company’s API or through its app,based on work led by the Hertie Schoo’s Center for Digital Governance. More here.

— Meta released new tools to stop so-called sextortion, which is the ability for bad actors to blackmail people — often minors — into handing over explicit photos of themselves. More here.

— In released court documents, Facebook created an internal program called “Project Ghostbusters” that used the company’s inside data for “intercepting and decrypting” app traffic from users of rival company Snapchat.

https://www.politico.eu/newsletter/digital-bridge/we-need-to-calm-down-about-ai-and-elections/

A document leaked yesterday by the news portal netzpolitik.org reveals for the first time which messenger and chat services are to be labelled as “high risk” according to the EU Council Presidency’s proposal of a Child Sexual Abuse Regulation. A “high risk” classification would allow issuing chat control bulk scanning orders to a provider, aimed at searching for potentially illegal chat content. The leaked document proposes that enabling “anonymous profiles or access without an account” should be considered a risk factor, as well as offering “encrypted messaging” or common interactive functionality such as direct messages, user posts or user comments. The document dates back to February, but an updated version from 10 April stipulates similar criteria.

“Privacy-friendly communication services such as Protonmail which can so far be used anonymously are to become the most extremely monitored services by imposing chat control scanning on them. So far securely encrypted messenger services such as Signal are to be turned into bugs in our pockets by imposing ‘client-side scanning’,” denounces Pirate Party Member of the European Parliament and shadow rapporteur Patrick Breyer. “EU governments want to go to war against the confidentiality and security of our digital communications as such. Everything that makes up the internet and digital communication, all of the modern reality of the internet is considered a ‘risk’ and to be combatted in the opinion of the EU interior ministers. We Pirates will not stop fighting for our fundamental right to digital privacy of correspondence and secure encryption – this is exactly why we digital freedom fighters are in the EU Parliament.”

https://www.patrick-breyer.de/en/leak-privacy-friendly-and-encrypted-messaging-services-are-to-be-penalised-with-chat-control-bulk-scanning-orders/

According to the latest draft text of the controversial EU Child Sexual Abuse Regulation proposal leaked by the French news organization Contexte, which the EU member states discussed, the EU interior ministers want to exempt professional accounts of staff of intelligence agencies, police and military from the envisaged scanning of chats and messages (Article 1 (2a)). The regulation should also not apply to “confidential information” such as professional secrets (Article 1 (2b)). The EU governments reject the idea that the new EU Child Protection Centre should support them in the prevention of child sexual abuse and develop best practices for prevention initiatives (Article 43(8)).

“The fact that the EU interior ministers want to exempt police officers, soldiers, intelligence officers and even themselves from chat control scanning proves that they know exactly just how unreliable and dangerous the snooping algorithms are that they want to unleash on us citizens,” commented Pirate Party MEP Patrick Breyer. “They seem to fear that even military secrets without any link to child sexual abuse could end up in the US at any time. The confidentiality of government communications is certainly important, but the same must apply to the protection of business and of course citizens communications, including the spaces that victims of abuse themselves need for secure exchanges and therapy. We know that most of the chats leaked by today’s voluntary snooping algorithms are of no relevance to the police, for example family photos or consensual sexting. It is outrageous that the EU interior ministers themselves do not want to suffer the consequences of the destruction of digital privacy of correspondence and secure encryption that they are imposing on us.”

“The promise that professional secrets should not be affected by chat control is a lie cast in paragraphs. No provider and no algorithm can know or determine whether a chat is being conducted with doctors, therapists, lawyers, defence lawyers, etc. so as to exempt it from chat control. Chat control inevitably threatens to leak intimate photos sent for medical purposes and trial documents sent for defending abuse victims.”

“It makes a mockery of the official goal of child protection that the EU interior ministers reject the development of best practices for preventing child sexual abuse. It couldn’t be clearer that the aim of this bill is China-style mass surveillance and not better protecting our children. Real child protection would require a systematic scientific evaluation and implementation of multidisciplinary prevention programmes, as well as Europe-wide standards and guidelines for criminal investigations into child abuse, including the identification of victims and the necessary technical means. None of this is planned by the EU interior ministers.”

The EU governments want to adopt the chat control bill by the beginning of June.

https://www.patrick-breyer.de/en/leak-eu-interior-ministers-want-to-exempt-themselves-from-chat-control-bulk-scanning-of-private-messages/