Google and Apple’s Anti-DMA Lobbying Strategy Goes All-in on Security and Privacy

I recently witnessed something quite remarkable at a privacy conference in Brussels. Freshly caffeinated, I sat down to listen to a panel on the role of operating systems in secure and privacy-preserving AI deployment. With agentic AI assistants such as Anthropic’s Claude Code and OpenAI’s Codex being connected to millions of people’s desktop operating systems every week, I was looking forward to hearing more about solutions to privacy and security challenges.

However, instead of a discussion about design choices and technical privacy and security protections, I found myself subjected to a passionate defense of Google’s right not to open its mobile operating system up to third parties’ AI systems. But it wasn’t Google calling out the European Commission for its proposal on Android AI interoperability measures, referring to them as “a privacy and security disaster” and “a policy incoherence with safety.” It was Apple.

Why might Apple, Google’s great rival in the battle for mobile operating system supremacy, so vociferously defend its biggest competitor?

Fear. Fear of a threat greater than Google’s competition. Fear of regulatory intervention that is forcing it, after a two-decade run, to open its infamously closed mobile ecosystem to third-party alternatives to the App Store and Siri. For this is what the Commission, under the Digital Markets Act (DMA), is moving towards.

Tech giants whose market dominance depends on tight control over closed platforms have responded by strategically leveraging privacy and security language as a shield against interventions that would dilute that control, giving users more choice and rivals a chance to compete.

Moving through the DMA’s gears: the EU’s pro-competitive interventions

Three years after its entry into force, the Commission’s application of the DMA is gaining momentum. This includes making use of some of the most promising tools at its disposal to prise open closed digital ecosystems: interoperability and data-sharing obligations.

In April, it proposed measures requiring Google to share anonymized search engine data with third parties under fair, reasonable, and non-discriminatory (FRAND) terms. This means sharing a significant part of its search dataset with other search engines and AI chatbots for the exclusive purpose of optimizing their own search functionalities. Allowing Google to share a meaningful portion of its search data will challenge its dominance, enabling competitor search engines and AI assistants to better compete with Google Search. With AI assistants representing a potential new form of internet search, giving rival AI developers access to this search data will also curtail Google’s ability to leverage exclusive access to its own search data to control the emerging AI market, as it is already doing.

This is part of the Commission’s broader, forward-looking approach to AI. It has already moved to prevent Meta from blocking competitors from offering their chatbots on WhatsApp, showing it has learned that failures to intervene early can be extremely costly.

But it is the DMA Android proceedings condemned by Apple that may be the most important of the Commission’s current enforcement actions. Recognizing the critical importance of operating system access to the functionality of agentic AI assistants, the Commission aims to prevent Google from leveraging its control over Android to dominate and shape the market for AI on mobile devices.

It hopes to achieve this by ensuring third parties offering competitors or alternatives to Google’s Gemini AI systems have a similar level of access to the Android operating system as Google does. This is achieved by requiring Google to open key Android system capabilities – currently only accessible to its own AI services – to rival developers. Competing assistants and apps will be able to interact deeply with the operating system, understand context across apps, and carry out actions such as sending emails, sharing photos, placing orders, or being activated by vocal commands on equal terms with Google’s own services.

The Commission is pushing for similar openness for iOS as it has demanded for Android. Apple, whose fortunes are particularly dependent on maintaining tight control over a deeply integrated proprietary ecosystem, is especially threatened by these types of interoperability obligations. Alongside AI, the Commission has pushed it to allow third-party app stores, enable non-Apple connected devices to access features previously limited to AirPods and Apple Watches, make file sharing via AirDrop interoperable, and much more.

The measures the Commission is looking to impose on Google and Apple are precisely what the DMA promised to do: open up monopolistic platforms controlled by Big Tech to competition and prevent gatekeepers from using their control over key digital platforms to constrain innovation or impose unfair terms on the businesses that depend on them. Without the interoperability mandated by the Commission, Google and Apple will automatically win the competition to provide the mobile AI assistant of choice for Android and iOS users. Competition, innovation and consumer choice are foreclosed from the outset by the unequal device and operating system access held by operating system gatekeepers.

The proposed measures open the AI market to genuine competition from rivals big and small, European and non-European, giving startups room to innovate rather than being squeezed out before they start. They should make it possible for users to choose their AI assistant from a differentiated range of products – privacy-focused, child-friendly, on-device, non-sycophantic, and so on – rather than being forced to accept Google’s or Apple’s offering.

Apple’s and Google’s lobbying campaigns

Given the threat these measures pose to their business models, Google and Apple are not going down without a fight. Instead, they have mounted a major, months-long lobbying campaign with one core argument: interoperability and data sharing are dangerous for consumers because they pose threats to their privacy and security.

On the sharing of search data, Google has sought to draw parallels to the 2006 release of AOL search data, which leaked the identity of hundreds of thousands of users. It also claims to have identified individuals from data anonymized following the Commission’s proposed methods within two hours. The AOL analogy, however, is questionable: in that case, the search data was released publicly and was not anonymized. The data Google will be required to share by the Commission will be anonymized and will only be made accessible to approved actors who are themselves subject to security and privacy standards. While technical concerns about the risks of de-anonymization should not be dismissed, Google’s conflict of interest on this topic means its claims should be taken with a hefty dose of skepticism.

At WWDC26, Apple announced it would not be releasing its latest Siri AI update in Europe, claiming the Commission had refused “to engage constructively on solutions that preserve privacy and security”. It noted that AI systems “can be hijacked to steal personal data [...] and to permanently alter files and account settings without a user's consent”. It did not, however, offer any answer to three obvious questions: why it permits third-party AI systems on macOS; why its own Siri AI is exempt from the same privacy and security concerns; and why iOS cannot be designed to give users granular control over AI assistant permissions. If it so chooses, Apple has the ability to design iOS in such a way that users have the power to decide which parts of their operating system, which apps, and which data any AI system – including Apple's own – is permitted to access.

Apple’s actions in delaying the release of Siri AI also fit a familiar pattern: it has spent years resisting attempts to open up its ecosystem, whether through third-party app stores, hardware interoperability, or now AI. It has made no secret of its dislike of the DMA, even calling for it to be repealed last year. Postponing the EU release of Siri AI may come from the same playbook as its delay in releasing live translation in AirPods to the EU, hoping consumer discontent will pressure the Commission into backing down. In the AirPods case, Apple ultimately relented, releasing withheld features to EU markets and delivering interoperability benefits to European businesses and consumers that would not have existed without the DMA.

When it comes to opening up operating systems, Google and Apple fall back on similar arguments. Both take advantage of the fear that security and privacy risks provoke in audiences, even when these are weakly substantiated. Their main arguments are that greater interoperability would weaken the security architecture of their platforms, create new avenues for data misuse, and force the exposure of sensitive system functions to third-party developers in ways that could not be fully controlled.

There is truth to the claim that a more open system will create additional opportunities for bad or negligent actors to exploit cybersecurity and privacy vulnerabilities. These risks should, however, not be overstated. Desktop operating systems, including Apple’s own macOS, have allowed users and developers to extend and modify system functionality for decades without resulting in the kind of systemic security breakdown Apple’s and Google’s lobbying implies will follow.

Closed systems are also not inherently more private or secure: Google’s and Apple’s mobile app stores have repeatedly been shown to admit malicious apps, including spyware and adware disguised as legitimate utilities. Cybersecurity vulnerabilities have also been identified on several occasions in Gemini. Closed distribution models, we should remember, are not themselves a guarantee of either security or privacy.

Even in more open environments, the vast majority of users tend to rely on trusted app stores, established brands, and default system protections rather than installing unknown or unverified software. In practice, security is shaped less by the theoretical openness of a system and more by the strength of its permission models, review processes, and standards. As such, interoperable systems can still be designed in a way that prioritizes safety, allowing secure and privacy-compliant services to emerge and naturally rise to prominence.

Privacy and security: legitimate concerns, convenient defense

Privacy and security are legitimate concerns, but tech giants appear to be taking advantage of this to defend their market power. Given the conflicts of interest in play, their objections cannot be the final word on how these trade-offs are resolved.

Interoperability and data-sharing obligations are a necessary component of any effort to open up markets that are centralized around a handful of dominant platforms. Genuine choice and competition in AI and beyond cannot happen without them, which is exactly why incumbents have every incentive to resist. That Apple and Google have mounted as intense a lobbying campaign as they have is itself a strong indicator of how much these obligations threaten their control.

Despite Google's and Apple's suggestions to the contrary, these obligations can coexist with privacy and security. Risks must be assessed, but transparently and independently by regulators and independent experts with appropriate technical expertise and without conflicting commercial interests. They should certainly not be resolved in favor of incumbents whose core interest is preserving the monopolistic status quo.