Why the Global AI Safety Agenda Cannot See African Harms

Rwanda's President Paul Kagame listens during the first-ever Global AI Summit for Africa where more than 1,000 policymakers, business leaders, and interest groups gathered in Kigali, Rwanda, Thursday, April 3, 2025. (AP Photo/Yuhi Irakiza)

Another Commission for Global AI Governance. How Can It Deliver for Africa?

A photo shows US Capitol in Washington, D.C., United States of America on July 2, 2026. (The Yomiuri Shimbun via AP Images)

Why Congress Must Codify a Chatbot Duty of Care

12 March 2026, North Rhine-Westphalia, Bergheim: Microsoft is building a data center region for AI applications in NRW. A total of three data centers are planned. The symbolic ground-breaking ceremony for the center in Bergheim was attended by Federal Minister of Digital Affairs Karsten Wildberger, NRW Minister of Economic Affairs Mona Neubaur, Head of Microsoft Germany Agnes Heftberger and Eoin Doherty (Microsoft). Photo by: Benjamin Westhoff/picture-alliance/dpa/AP Images

Microsoft is building a data center region for AI applications in Bergheim, North Rhine-Westphalia, Germany. Photo by: Benjamin Westhoff/picture-alliance/dpa/AP Images

The Case for a Temporary Moratorium on Large Data Centers in Europe

Google CEO Sundar Pichai speaks at the Singapore FinTech Festival in December 2020. (Source)

Google CEO Sundar Pichai during a meeting at the European Commission in Brussels, Belgium. (

Google Android Antitrust Ruling Shows the Next Battle is Over Who Controls Trust

How the AI Deepfake Boom is Outpacing Europe's Safeguards

High-level Meeting to launch the Global Dialogue on AI Governance. UN photo/Laura Jarriel

High-level Meeting to launch the Global Dialogue on AI Governance. UN 

From the AI Accountability Gap to a Global Accountability Infrastructure

US Secretary of State Marco Rubio attends the East Asia Summit Foreign Ministers' Meeting, part of the Association of the Southeast Asian Nations (ASEAN) Foreign Ministers' Meeting in Pasay, Metro Manila, Philippines Thursday, July 23, 2026. (Brendan Smialowski/Pool Photo via AP)

Trump Cannot Use Immigration Law to Silence Technology Researchers

A gowned worker adjusts a test coupon for indium phosphide wafer certification testing in a Coherent manufacturing facility where lasers are developed for transmitting data among chips, which could decrease power use by AI systems, on Tuesday, June 16, 2026, in Sherman, Texas. (AP Photo/Jeffrey McWhorter)

The EU's AI Boom Could Undermine its Own Chip Strategy

Woman defending herself against violence, illustration of a giant fist coming down. 

A Roadmap for Confronting the Chilling Effects of Censorship, Surveillance and New Technology

A magnifying glass hovering over the OpenAI logo in a web browser.

OpenAI logo with magnifying glass, CC BY 2.0, Wikimedia Commons (by Jernej Furman).

Despots, DSA Fines, and an 'Autonomous' AI Attack

Open AI CEO Sam Altman speaks to reporters after meeting with Sen. Bernie Sanders (I-VT) in the Dirksen Senate Office Building on Capitol Hill on June 3, 2026 in Washington, DC. (Photo by Chip Somodevilla/Getty Images)

How the OpenAI-Hugging Face Hack May Affect the Geopolitics of AI Governance

People visit the booth of Kimi, an LLM developed by the Chinese startup Moonshot, during the World AI Conference in Shanghai, China, Monday, July 20, 2026. (FeatureChina via AP Images)

‘Open-Washing’ Is Everywhere in AI. Four Criteria Cut Through It

Michael Kratsios, Director, White House Office of Science and Technology Policy, speaking at a hearing of the House Committee on Science, Space, and Technology at the US Capitol in Washington, DC. (Photo by Michael Brochstein/Sipa via AP Images)

OSTP Vision for Science is a ‘Golden Age’ for Corporations

Close up of the Ray-Ban Meta Wayfarer camera integrated into the smart glasses frame. Shutterstock

Close up of the Ray-Ban Meta Wayfarer camera integrated into the smart glasses frame. 

An Indian Trans Person’s Harrowing Encounter Fuels Backlash Against Meta AI ‘Pervert Glasses’

Dormont Borough, PA. (Photo credit: Local Government Academy)

AI Reached Local Governments Before Policy Did

French President Emmanuel Macron welcomes European Commission President Ursula von der Leyen before a meeting at The Elysee Presidential Palace in Paris on July 22, 2026. Photo by Raphael Lafargue/Abaca/Sipa USA(Sipa via AP Images)

What France’s Under-15 Social Media Ban Means for Europe

The image shows a colorful, abstract representation of a human profile, symbolizing neurodiversity. The person’s head and neck are composed of interlocking jigsaw puzzle pieces in various bright colors, such as red, yellow, green, blue, and orange. The puzzle pieces forming the head extend outward, suggesting complexity and diversity within the brain. The use of different colors highlights individuality and the unique ways people's minds work. The background is a solid blue, which contrasts with the vibrant colors of the puzzle pieces. The image reflects themes of mental health, cognitive differences, and embracing neurodiverse thinking.

China’s First Commercial Brain Implant Shows Why Neurotechnology Needs Governance

New Delhi: Cockroach Janta Party (CJP) supporters wave the national flag during a protest demanding the resignation of Union Education Minister Dharmendra Pradhan over NEET paper leak issue, in New Delhi, Thursday, July 23, 2026. (Salman Ali/PTI via AP)

India's CJP Protests Meet Internet Shutdowns and Pervasive Surveillance

A blurred black-and-white portrait of a man on the left, with the xAI company logo displayed on a screen to the right against a black background.

What Data Was Your AI Trained On? xAI Doesn't Want You To Know

From coordinated harassment campaigns to sextortion and digital violence, survivors reveal how AI-generated sexualized deepfakes are reshaping abuse while regulation struggles to catch up.

This article discusses real stories of non-consensual sexualized deepfakes, image-based abuse, online harassment, sextortion, and the psychological impact of digital violence.

After being targeted with a swarm of fake nude photos of herself, Julie starts to have intrusive thoughts. Every time the 31-year-old journalist living in Rio de Janeiro enters a lift full of men, her thoughts start racing. Have they seen the photos? Have they had access to an intimate version of her body she never meant the world to see? The walls seem to come closer as an even scarier thought creeps in: what if one of the men was the creator of the fake images?

Julie no longer feels safe anywhere. And she feels as if she lost control over her body. What started as an online attack has now spilled into the offline world, affecting her everyday life, her reality.

On the other side of the world, in the UK, Laura* scans the Internet for fake photos of herself. Laura is a seasoned sex worker. She was alerted that her content might have been leaked from sex work marketplace websites that only paying members can access. On Google, she stumbles across a picture of herself kissing somebody famous.

The scenario is fake – the kiss never happened.

She feels sick to her stomach. As a sex worker, she is used to uploading lewd photos of herself to the Internet. But running into images that she knows have been altered feels like a betrayal. Just like Julie, Laura suddenly became aware that she had lost control over her own body and her own image.

Both Julie and Laura have fallen victim to a technology that uses AI to create fake pictures: ”Nudification” apps, also known as non-consensual sexualisation tools (NSTs), became known to the mainstream media after it was revealed that X’s AI-powered chatbot 

 had been used to create sexualised images from real photographs, mostly of women, but also children and members of the LGBTQI+ community. The backlash was so intense, it eventually resulted in a ban on such tools being included in the AI Omnibus – 

the law promising to “simplify” digital regulation in Europe

You see a stranger, you take a photo and almost automatically, you are able to strip their clothes away – as easily as snapping your fingers. Essentially, that is the power of an NST. The process behind it is more complex, though: an enormous amount of data is required to train the generative AI models that power these tools, which basically work by remixing existing photos of nude bodies to create an image that matches the face and outline in the photograph provided – just without the clothes.

For sex workers, the technology offers an additional insidious harm: by creating fake nude or sexualised images of sex workers, people can have access to the services they sell without ever paying – and with images created by ripping off other sex workers’ labor.

While Grok has so far been the most prominent case of AI-generated mass-production of fake sexualized images, plenty of other apps are able to do the same. The idea precedes them, of course: For a long time, photo montages of celebrities or private individuals superimposed onto photos of porn stars or naked models were commonplace. The difference now: AI allows any user to generate extremely realistic deepfakes that are incredibly difficult to distinguish from real photos.

Kolina Koltai, a senior researcher at the investigative journalism group Bellingcat from the Netherlands, has extensively looked into this kind of tool. She warns that “the speed and the ease with which someone can now realistically create content makes the difference. Nowadays, you don’t need money or a technological skill-set to make it happen,” she adds.

With the European Commission and national regulators putting a spotlight on the issue, the impact on sex workers this new wave of imprudence is producing has hardly been addressed. With old montages, people rarely talked about the women whose images had been “decapitated” to create manipulated versions. Now, porn websites are scraped without permission. Not only does this constitute a form of digital violence and abuse, but also theft; for sex workers, their content is their livelihood. But typically, they are not afforded victim status: “People unfortunately seem to be less sympathetic towards sex workers,” says Koltai.

While digital violence against women is not a new phenomenon, Julie sees some worrying changes: she recounts many attacks against her in the form of hateful replies and comments. The difference now is that these are accompanied by extremely vivid nude images of herself. Previously, the attacks originated in her having “strong opinions” on politics that she expressed on her platforms. However, after the COVID pandemic, she noticed a shift: the targeted attacks became more 

Julie was one of the victims of the campaign of Grok-generated nude images in early 2026, being swarmed by many different profiles on X. She would open the replies on her profile and see herself in sexualised poses in nude photos that looked eerily realistic.

She could tell that the accounts reposting the pictures had been created in different countries. They all followed the same accounts and seemed like a coordinated group of “ghost profiles” without any individual, real users behind them: “Only three of the more than 20 users that attacked me had over a thousand followers. It seemed like an interconnected network: they had a fixed speech and seemed to already know what they were going to say when they discovered my page.”

Celebrities and public figures are often the easiest targets of the swarm that attacked Julie: Georgia Meloni, Italy’s Prime Minister, 

took it upon herself to debunk a fake photo of her posing on a bed in a chemise.

 While the photo is far from pornographic, it was enough to cause public uproar and a dent in her reputation.

The deputy Prime Minister of Sweden, Ebba Busch, 

. Users on X instructed Grok to undress her in photos, augment her breasts and buttocks, or change her position so that she would seem to bend over in a sexualised posture. Busch also responded publicly to the scandal.

For Koltai, there is a monetary incentive to keep the abuse going: “Generally the people behind these apps do it because it makes money for them. It is an incredibly profitable business.” 

A report by the Tech Transparency Project estimates that NSTs have been downloaded a whopping 483 million times only on the Apple Store, and made more than $122 million in lifetime revenue.

Targeted attacks: When the danger is closer

Being swarmed by strangers with pornographic deepfakes and hateful messages is not the only way this violence manifests. “In many cases, the perpetrator is someone known to the victim,” Koltai explains. In Germany, 

the harrowing case of Collien Fernandez caused a country-wide uproar

: For many years, her husband – a prominent comedian and actor – used both real and AI-generated photos of her as a form of sexual abuse, impersonating her on online platforms to contact strangers.

For sex workers, this reality of violence is even more complex, and contrary to what many people may assume, women are not the only target.

Kelly* is a non-binary refugee assigned male at birth (AMAB) who lives in Turkey and carried out sex work for a period of time to support themself while settling in a new country. They were careful to ensure that no photos of them used in online advertising would give hints revealing their identity. As a refugee sex worker, they might face sanctions both by the Turkish government and international refugee authorities.

None of that mattered to a colleague whom Kelly met in the refugee community. Aware of Kelly’s dabbling in sex work, the two ended up having a discussion over politics. “It was so stupid,” Kelly recalls.

A few days later, Kelly received photos of themself – completely nude and with their genitals on display. Their colleague was threatening to report them to the authorities. Kelly was terrified. Other people in the community were said to have received similar threats. Kelly never knew what app was used to generate the images.

Annie*, a non-binary sex worker from Poland, assigned female at birth (AFAB), also saw themself become a victim of sextortion based on AI-generated content (an extortion based on intimate images). Their ex-girlfriend sent artificial “nudified” pictures of Annie to their family, and exposed them as a sex worker. The images also spread online, and because the original photos showed parts of the house and neighborhood they lived in, their address was eventually revealed. They were forced to move and endured an unstable housing situation for months. A year later, while describing the case, Annie is still traumatized.

For sex workers who often live under precarious conditions due to the criminalization and stigmatization of their livelihood, the constant threat of being exposed is an additional burden. All workers interviewed as part of this investigation reported a loss of income since they started to fear advertising for themselves in their regular channels. Annie recalls being scared of working for months.

In this particular case, seeking law enforcement – or even using the limited redress mechanisms offered by platforms – is not always a viable option. In most cases, platforms ask for identifying information to take down content, which can put sex workers at risk of being exposed.

But even for those who are not sex workers and don’t face such restrictions, seeking the help of platforms can be equally complicated.

Initially, Julie felt like she should step away from the computer and wait for the whole thing to blow over. But the attacks continued. So she decided to take matters into her own hands and push back. Contacting the police and lawyers, she recalls as an incredibly frustrating experience: “They are not equipped to deal with it. The police were basically acting like I was overreacting because the photos were fake.”

Thousands of kilometers away in the UK, Laura was given a similar response. The police said they could not do much since the photo was AI-generated. Therefore, there was no way of knowing whether it was really her in the images.

Cordelia Moore, a Digital Violence Consultant in Germany working directly with survivors, explains that in many cases, victims are not even aware that they can report their experience to the authorities: “With synthetic images, there is still a large gray area: law enforcement is not prepared to deal with such cases.”

And within the platforms, it seems like that is also the case. When Julie reported the photos to X, they were taken down – but there was no trail left to follow the offenders, as many had used ghost accounts or bots. X did not provide a feasible way to properly track perpetrators.

This experience seems to be common. Lejla Podgoršek, Senior Policy Manager at AlgorithmWatch who has also been investigating the apps, explains how she has seen accounts on X that were reported for using Grok to create sexualized deepfakes disappear temporarily only to later come back to the platform.

Beyond big players like Grok, the market for nudification apps can get even murkier. “Many of them are not listed as obvious nudification apps,” Lejla explains. “At first glance, they look like something different – like a romantic companion app, for instance. It takes closer examination to see what exactly they provide, which makes it difficult to keep track.”

”Most apps are shared in closed forums. X got in trouble because Grok is more public,” Koltai from Bellingcat explains, before adding: “Grok is the world's most popular nudification service – even if initially it hadn’t been built for that.”

Laura found very little sympathy for her ordeal, even if it had lasting consequences in her life and her work. “People would say: Well, what did you expect, you post this kind of content online,” she recalls.

However, private individuals seem to hear exactly the same. Julie recalls friends using the same arguments to blame her for the deepfake attacks: “Well, you posted pictures of yourself online, what did you expect?”

The goalpost is ever shifting for those at the receiving end of the violence. This unites most victims of non-consensual nudification, whether they are public figures, private persons, or sex workers.

All of this happens in an era that pushes people to have an online presence, and in which working online is still the safest option for many sex workers – to be able to screen and vet clients and have more control over how, when, and where they work. In fact, different groups among those affected have more in common than might transpire at first glance.

Graphic by Ana Ornelas, based on the analysis of 25 victim accounts.

Oftentimes when non-consensual deepfakes are discussed, the language used suggests that they represent a recent issue. While the realism and level of low friction and ubiquity of undressing apps is certainly a novelty, the problem itself is not.

In a seeming attempt to make everything look like a harmless joke, Elon Musk, the owner and CEO of X, 

used Grok to undress and put himself in a bikini

. However, Musk was never risking his reputation, his relationships, or his job by appearing in a bikini. Plus, the trillionaire did this voluntarily.

This is perhaps the most fundamental and insidious aspect of NSTs and non-consensual intimate image sharing: Sexuality and nudity are used as a weapon to coerce, control, attack, abuse, and blackmail.

While such a practice could also be carried out purely for sexual gratification, this investigation shows that this is seldom the case. For the most part, sexualised images are created with the intention to humiliate and avenge. As Julie puts it: All the perpetrators had to do in her case was to write a prompt to Grok and press a button. For her, this simple action caused a devastating wave whose rippling consequences will affect her forever.

After the abuse happened, Julie felt that her perception of others has been altered: “Nowadays, I perceive men as always trying to cross my boundaries, online and offline. I feel like I am constantly under surveillance.”

Feminist historian Gerda Lerner writes that “the class position of women became consolidated and actualized through their sexual relationships”. The issue is that the sexuality of women and gender-expansive people is used as a weapon to inflict violence.

This can only happen in a society that accepts the policing of bodies and sexuality to destroy the reputation and relationships of individuals. The “loss of control” over their bodies is something all survivors describe. But Julie goes even further: “I realized that I never really had it.”

She also feels that the female body is “public property,” which relates to what feminist historian Silvia Federici wrote in 

: When peasants lost access to common land, the trade-off made by the patriarchy was that the female body became the new commons.

The misogyny that underlies and enables this kind of abuse is the same misogyny that shields perpetrators from accountability while blaming victims for the violence they endure.

NSTs function as a mirror of how this operates more broadly in society: women and gender-expansive people are disproportionately targeted by violence, with sex workers facing even greater vulnerability because of stigma, criminalization, and marginalization. When that violence is exposed, victims are blamed, while any association with sex work (or even behaviors merely perceived as adjacent to it) is weaponized to further discredit their testimony.

And as long as this misogyny remains unaddressed, it will continue to ripple through every new technology that emerges.

*Names have been altered to protect the identity of the victims.

This investigation is supported by the Algorithmic Accountability Reporting Fellowship by AlgorithmWatch.

 to Europe’s technological sovereignty. Yet, it 

 Europe needs to triple data center capacity, nor how existing infrastructure is being used, nor how rapid expansion can be reconciled with Europe’s climate, energy and democratic commitments.

solid evidence for the need to triple data center capacity

, with evidence to suggest that many data centers in Europe 

ew data center development will be dominated by US tech and co-location firms

 who prioritize serving the largest multinational corporate cloud providers, most of which are US-based.

Nevertheless, the European Commission is aggressively dismantling environmental protections and democratic oversight to achieve this target. We believe that instead of scrambling behind superpowers in a race with no clear finish line, Europe needs to make clear-headed decisions backed by solid evidence. Europe should not rush into an infrastructure race simply because other global powers are doing so. It should make evidence-based decisions about its actual digital needs and pursue an industrial strategy that balances technological ambitions with environmental limits, social impacts, jobs and workers' rights.

Until these questions are answered, we believe Europe should pause the construction of new large-scale data centers. A Europe-wide moratorium would create the space to assess whether this expansion is necessary, sustainable and in the public interest.

1. Exacerbating the cost of living crisis for Europeans everywhere

Evidence supporting the need for data center expansion for economic growth is scarce. We do, however, have examples from across Europe of how data centers make life more expensive for residents, not less.

Take Ireland. The growing number of data centers in Ireland last year used

added hundreds of euros to household electricity bills

 (a cumulative average of €360 between 2015 and 2023) and drained 

0.1% of the Irish workforce (3,300 people)

This pattern is set to be replicated across Europe. In the Netherlands, much-needed housing and 

50 schools and daycares could not be built 

because a large data center got priority access to the electricity grid. Meanwhile, the Dutch national high-voltage grid operator, TenneT, 

 of data centers are threatening the security of the grid.

One of Spain’s largest regions is constructing data centers that would consume 

 than all other current economic activities combined. These data centers would occupy up to 32% of the land set aside for renewable energy generation—

energy that would not be added to the grid

 but rather exclusively allocated to powering these facilities. Furthermore, contractual clauses give these data centers priority access to water in areas already grappling with 

. As a result, energy and water gentrification is occurring, 

 in the coming years, adding 28.6 GW of demand to the grid — the equivalent of 28 nuclear power plants. A 

scientific report by the national NGO the Shift Project

 shows that this directly conflicts with the need for electrification in several sectors in line with the green transition. In Marseille, 

the growth of data centers has put a halt on the electrification of the port and its associated ferries

, delaying much-needed transition away from fossil fuels and marine pollution.

Overall, Europe’s acceleration plan only serves to heighten tensions between the economic areas that benefit from data centers, and those home to the most vulnerable that are turned into sacrificial zones.

2. Lock in dependencies on US tech power and fossil fuels

Europe’s AI sovereignty problem runs far deeper than frontier access

. The current “AI First” approach in Europe means 

. This lock-in further entrenches dependencies on US tech power at the expense of European industries, households, schools and hospitals. Large data centers compete for limited energy and water resources while our governments allow them to burn fossil fuels with impunity as they build new fossil gas plants, 

, and extract our data and expertise without fair compensation.

The data center industry openly admitted that the only way we can triple our data center capacity is by burning fossil fuels, effectively 

”choosing between AI and climate goals”

 while Europe experiences historic heatwaves and is the 

 on cloud computing run by the three US tech giants. It also imports more than €400 billion in fossil fuels from the US as well as Russia and the United Arab Emirates. The energy demands from data centers and the associated climate and environmental costs are significant, yet poorly addressed.

3. Sacrificing Europe’s democracy and climate goals

Europe already has the Energy Efficiency Directive, intended to reduce emissions across the EU by 55% by 2030. However, growing evidence shows that expansion of data centers is incompatible with this target, and 

 with the Directive causes a significant regulatory blind spot.

In Ireland, agencies have stated that the pace and scale of data center growth 

“will make it very challenging to meet both the EED and RED targets.”

 Energy demand from data centers in Ireland grew 

. During the same time period, demand from other sectors grew by only 2.8%. Even in the most ambitious scenarios, Ireland could

, with all gains in wind energy in the country between 2017 and 2023 

outstripped by growth in data center demand

. The country’s Climate Change Advisory Council has gone so far as to state that rapid growth from data centers is 

To triple data center capacity, the EU will force member states to build “acceleration zones” yet allow data centers to continue operating in secrecy, in a move overriding basic democratic principles. Existing data centers 

 with minimal transparency requirements, and instead of tightening enforcement, the EU has 

succumbed to US tech lobbyists who successfully pushed to classify all individual information about data centers as confidential.

 Decision-makers, journalists and citizens cannot access numbers about individual data centers through freedom of information requests.

 shows, over the next six years, Europe is on track to “drive a 121 million-tonne surge in CO2 emissions—nearly equivalent to the total emissions from all gas power plants in Italy, Germany, and the UK in 2024 combined,” as data center power demand soars in the EU.

unable or unwilling to gather basic information

 about the effects of its expansion plan, nor enforce compliance with its Directive about how existing infrastructure operates, how can the EU have an evidence-based approach to policy? How will Europe meet its climate goals if it allows data centers to burn fossil fuels on-site, outstrip the growth of renewables and claim priority access to our grids? What is the implication of this on the electrification of other sectors?

European-wide moratorium on large data centers

Fast-scaling computing is breaking non-scalable infrastructures all across Europe. The return on investment 

, but the real costs to households, local industries and our natural environment are high and the employment prospects low. Meanwhile, every new data center intensifies a global supply chain of extractivism. When Europe embarks on a race to build more and more data centers without proper justification or democratic deliberation, we validate an exploitative model that hurts poorer regions in Europe and vulnerable communities across the entire planet.

The EU argues it has to create the demand for AI to justify the data center buildout. Yet building bigger and faster will primarily benefit US tech power—while further entrenching Europe’s digital and energy dependencies and sacrificing its commitments to democracy and decarbonization. US tech power 

 to impose their deregulation agenda. Data center operators prey on poor and marginalized communities to force the fast deployment of data centers before any meaningful contestation can be made.

While we criticize Europe’s reliance on US tech power, simply replicating the same models but with a European stamp is not enough and would actively harm communities and undermine broader development. Instead, a break from the extractive, bloated and financially risky models of the past is needed.

Europeans deserve clear answers before more large data centers are approved. The Commission should demonstrate how much additional capacity is needed, who will benefit, how projects will avoid environmental harm, how data center buildout is compatible with legally binding climate commitments, how they will support new renewable generation rather than compete for existing supplies, and how households will be protected from higher costs. 

Until then, Europe should adopt a moratorium on new large-scale data centers.


After eight years of litigation, Europe’s highest court has dismissed Google’s appeal, upholding the European Commission’s finding that Google used its control over Android to make its own search engine harder to avoid. But while that legal battle focused on contracts and market dominance, Google is now looking to shape how trust is established across the Android ecosystem.

On July 2, the Court of Justice of the European Union delivered its 

 in the Google Android antitrust case. The ruling made the penalty imposed by the General Court definitive. One of the most consequential technology competition disputes of the past decade, the case examined whether Google broke the law by using contractual restrictions, including anti-forking agreements, revenue-sharing arrangements, and tying practices, to preserve its dominance over the Android ecosystem and mobile search. The Court dismissed Google's appeal.

Yet over these same eight years, Google has shifted how it maintains gatekeeper control. The latest iteration is the company's 

: starting in September 2026, developers worldwide who wish to distribute apps to certified Android devices will be required to agree to it.

The timing is striking. While the Android antitrust case asked whether dominant platforms could preserve their power through contracts, Google's developer verification system points to something deeper: the next generation of platform control may operate through governance of trust itself.

The pattern is familiar. Whenever a single firm has exercised control over a technological ecosystem, it has argued that greater openness threatens security. In the United States, the 

 argued that allowing third-party devices to connect to the telephone network would compromise its integrity and reliability. In the landmark 1968 

 the Federal Communications Commission (FCC) rejected a blanket prohibition on third-party devices, establishing that users could attach non-Bell devices as long as they did not harm the network. In 2007, 

Skype requested the US telecom regulator to apply the Carterfone case to the smartphone industry

to prohibit telecom operators from restricting end-users’ freedom of choice regarding smartphones. This marked the beginning of net neutrality.

In 2004, Microsoft argued before European competition authorities that disclosing interoperability information would undermine Windows' security architecture. The European bodies 

rejected security as a blanket justification

 for refusing interoperability, even while acknowledging the underlying concerns were legitimate.

Today, similar arguments continue to shape digital markets. Apple argues that the unfettered installation of apps (sideloading) and interoperability obligations under the Digital Markets Act (DMA) threaten the integrity and security of its systems. Before the Court of Justice of the European Union, Apple has argued that 

the DMA's interoperability obligations raise questions of proportionality

With its developer verification, Google argues that unrestricted app distribution itself poses unacceptable risks to Android users. The use of 

 to preserve control over technological systems is therefore hardly new and still as incorrect as it was in the past. The underlying security concerns may be real, but they do not necessarily justify centralized control by the incumbent where less restrictive and more interoperable alternatives are available.

Keeping control over the Android ecosystem: from contracts to trust

Android’s success was built on openness. Launched in 2007 as an open-source operating system, it offered manufacturers, developers and users more flexibility than competing platforms. The European Commission’s 

later revealed the tension at the heart of Google’s strategy, which sought to maintain control over an ecosystem whose appeal depended on appearing open.

The investigation culminated in a landmark 2018 decision finding Google in violation of EU competition law. Google appealed to the Court of Justice of the EU, which recognized that digital ecosystems cannot be understood through traditional market definitions alone. They require a "multi-level" and "multi-directional" analysis, since products, services, developers, manufacturers, and consumers interact across interconnected platforms.

That case closed only weeks ago, but it wasn't the only one before the Courts in recent years. Another case, involving the Android Auto app, shows how Google increasingly relies on ecosystem and interoperability control rather than contracts alone. 

The CJEU examined interoperability questions

 referred by an Italian court in the dispute over an EV-charging app. Google had refused to make the app interoperable with Android Auto, citing security concerns and the absence of a specific template for EV-charging apps, since the available templates were limited to categories such as messaging and media apps.

The Court found that although Google hadn't prohibited the app's distribution, control over competitors' participation in a digital ecosystem can itself be a source of market power: decisions about who may access a platform, when, and under what technical conditions can shape competition as much as contracts do.

Together, the two cases illustrate different generations of platform control:

control through contracts (the Android antitrust case)

control through ecosystem governance and interoperability (the Android Auto case)

Android revolutionized the mobile market by being under a Free Software license: anyone could run, install, distribute, and develop software, including alternative versions of Android itself. This vision of Android as a general-purpose computing platform is reflected in initiatives such as the Free Software Foundation Europe's 

Latest

Podcast

Recent

Our content delivered to your inbox

Thank you!

Around the globe