Trust vs. Anti-Trust

This week, the US Department of Justice along with sixteen states filed a significant antitrust case against Apple. The complaint accuses Apple of having a monopoly in high-end smartphones and employing five tactics to maintain that monopoly. Apple has supported these tactics with the rationale that prioritizing privacy and security makes the iPhone a trusted platform. The government disagrees, however, and claims Apple is effectively blocking competition and restricting choice and innovation for consumers in its antitrust lawsuit.

In this essay, I’ll cover the basics of the complaint and then speculate about how to think about these issues in the emerging age of AI.

Having a monopoly is not illegal, and neither is anticompetitive behavior. But having a monopoly and maintaining that monopoly through anticompetitive behavior is. The first step in the government’s case will be proving that Apple has a monopoly.

In the late 1990s, as part of my job as a Microsoft analyst at Morgan Stanley, I watched the Microsoft antitrust trial from within the courtroom. I remember the effort the government had to put in to prove that Microsoft’s 95% market share in operating systems represented a monopoly. It seems to me that the government has a much harder challenge, proving that Apple’s 65% marketshare in high-end smartphones represents a monopoly—especially when its marketshare is only 23% worldwide.

The government then has to prove that Apple maintains that monopoly through anticompetitive behavior. The complaint lists five accusations:

1. Apple suppresses so-called “super apps.” The claim is that Apple prohibits super apps like WeChat on the iPhone because these apps would diminish Apple’s monopoly since users could simply use the super app and its sub-apps rather than using Apple software and App Store apps. This feels like an odd complaint to me given that 1) the primary super app in the world is China’s WeChat and the US government is considering banning the other great Chinese app, TikTok and 2) I enjoy a full suite of apps on my iPhone today. For instance, when using Instagram, I get notifications from Facebook, Facebook Marketplace, Facebook Messenger, and Threads (I’d likely also get notifications from WhatsApp if I allowed it more access to my phone). I have similar inter and intra app experiences from my suite of Google tools as well. Yes, these tool suites do not have the complete functionality of WeChat but they provide an optional ecosystem to send messages and email, post and consume content on social media, consume media, manage photos, manage home devices, buy and sell items, create documents, etc.
2. Apple suppresses cloud streaming games to support its own gaming and to encourage customers to buy more powerful phones. I’m not a gamer, so don’t have any direct experience with this issue. Apple’s defense of this practice has primarily been about protecting users by requiring all app updates to go through an approval process (streaming games can update multiple times a day). All that said, Apple has already made some concessions on this point in the EU, so it seems like an issue that may be easily resolved.
3. Apple degrades the messaging experience by highlighting non-iMessage texts in green rather than blue. Before you dismiss this and say, “The government is really suing Apple over the green bubble issue?”, this has proved to be a material advantage for Apple. People don’t like being the green-bubble-Android-poor-cousin in group texts and buy iPhone’s just to avoid the stigma. That said, I’m not sure what the government’s practical remedy might be. As an iPhone user, I want to know which texts are encrypted (iMessages) and which are not (all others) so the distinction between blue and green makes sense. The complaint seems to imply that the government finds fault with Apple for not having an Android version of iMessage—are they going to attempt to force Apple to port iMessage to Android? Wouldn’t that create the possibility of an iMessage monopoly? As with cloud streaming, Apple has already announced they will start supporting RCS, an independent messaging protocol that may remove much of this issue.
4. Apple suppresses competition in smartwatches by restricting access to APIs that allow certain functions between the Watch and iPhone. I can see how this argument might hold if a) the iPhone is determined to be a monopoly and b) the government can prove that the Watch is an extension of the iPhone and Apple is tying one to the other. That said, as a user, I appreciate Apple’s work to synchronize things like messaging across all my devices and I imagine this would be much harder to do with a fleet of third-party devices. That means that opening the smartwatch APIs might increase competition, but might also degrade my experience.
5. Apple suppresses competition in digital payments by restricting access to the chips that allow touch payments through Apple Pay. Apple has already conceded this point in the EU and will be opening up to competition, likely meaning this could be easily resolved in the US as well.

What this might mean in an AI future.

Much of the question about antitrust in tech revolves around how you define product boundaries. Was the browser part of the Microsoft OS, or was it a separate product that Microsoft “tied” to Windows? Are the APIs that support iMessage part of iOS? If yes, then those technologies are part of the (possible) monopoly, not necessarily tools used to maintain that (possible) monopoly.

We think AI will significantly disrupt today’s product boundaries. Foundation models will provide baseline support for new categories of applications. We call foundation models the “aiOS” because of the potential for them to create new platforms—and possibly new monopolies. How will these new foundations alter the boundaries and (possible) monopolistic positions of current OS players?

We are obsessed with the ability for AI to write code, especially dynamically. Google has provided a peak into the future through its dynamic design demo, showing a generative system coding a custom interface based on a user conversation. How might that change the definition of apps? What if these dynamically designed apps are one-use only and disappear when the task is done? If these new dynamic apps draw on a suite of tools and models provided by an OS vendor, does that OS vendor need to allow access to all tools and models, or can they keep some of those to themselves?

We are also obsessed with the potential for AI to act as our agents. We’re both excited by this capability and fearful of it. Having an AI agent that understands me well and can take the actions I want it to could be immensely helpful. But, in order to be truly helpful, this agent would need to know quite a lot of my private information, have access to all my accounts, and be able to to make transactions, send messages, and a variety of other things on my behalf.

If I were to delegate agency like this to an AI, I would need to trust this AI more than I trust my technology today. At the heart of this antitrust case is the tension between Apple’s methods of building trust by constraining parts of its ecosystem and the innovation potential that might appear from others if Apple was more open. This tension will only get more pronounced in an AI future, as the innovation that we hope for will require more constraints to maintain privacy and security.

While this future may still be years in the future, it may arrive before this antitrust case has gone through all of its appeals. Given the detailed nature of the current complaint, any possible resolution may not have an impact on how AI rolls out. But, successfully defining the iPhone as a monopoly might create a precedent that impacts the future of AI. We know that a closed system has privacy and security advantages. The government is saying that Apple’s closed system is anticompetitive. Will it force Apple to open its system, potentially exposing additional privacy and security risks?

Unfortunately, regulatory failure to provide mechanisms and infrastructure for personal data protection has left the question of data privacy and security to Big Tech. I bet that a significant portion of iPhone customers opted into Apple’s tight ecosystem in part because of privacy and security. If I’m right, that means that Apple’s market share and potential monopoly is, in part, due to the closed system effects that the government finds anticompetitive.

I agree with the government that antitrust is important, and preventing anticompetitive behavior at monopolies is essential to a functioning innovation economy. That said, I would hate to see data privacy and security diminished just as we are trying to take advantage of AI’s power to be a personalized digital partner. And, like it or not, closed systems like Apple’s can provide the best data privacy and security.