It’s outrageous for netizens to use AI to build computers this time: Redmi can run iOS and Mac and run Android

Successfully running iOS 16 on Redmi and running Android apps on Mac, these attempts have suddenly become popular in the past month or two.

Not long ago, the owner of Bilibili UP even successfully made the OnePlus Ace 5 run Hongmeng 6. He also repaired the GPU driver and was able to display the desktop, system animations and application windows normally.

Almost at the same time, someone also successfully lifted the computing performance limitations of some NVIDIA RTX 20, 30, and 40 series graphics cards by modifying the underlying configuration. The RTX 4090’s throughput rate even nearly doubled in specific tensor calculation tests.

What’s even more outrageous is yet to come.Macs run Android applications, iPhones run Android games, and even Switch games are ported to Windows, Mac, and iPhone, which have successively become new challenges for developers. Recently, someone even showed footage of Switch exclusive games “The Legend of Zelda: Tears of the Kingdom” and “Mario Kart World” running natively on Mac.

Of course, the degree of completion of these achievements varies, and some projects are actually still in the technical verification stage. But compared to the past, which required a lot of professional knowledge, time and energy to complete system transplantation and reverse engineering, individual developers now have the opportunity to challenge complex projects that were originally difficult to achieve with the help of AI Coding Agent.

Engagement is nothing new. But today, when mobile phones and computers are becoming more and more mature, and rooting and jailbreaking are gradually fading out of the public eye, why can AI make the gaming industry lively again?When more and more people try to break through the software and hardware limitations set by manufacturers, what changes will it bring to the consumer electronics industry?

Working on machines is not new, but the way AI works is very different

Having experienced the early development of smartphones, it should not be difficult to understand the appeal of mobile phone culture.

From flashing third-party ROMs to Android phones, to jailbreaking iPhones, and hacking Apple on PCs, over the past decade or so, some developers and users have been keen to challenge the boundaries of software and hardware preset by manufacturers. Some are for new features, some are for improved performance and experience, and some are purely to prove what’s technically possible.

However, as smartphones become increasingly mature and manufacturers continue to strengthen system security and ecological control, the influence of third-party ROMs gradually declines. A large number of functions that in the past required Root and jailbreak have also been absorbed by the native system.

But AI Coding has changed the situation.

Take the Redmi Note 10 Pro running the iOS 16 open source project ios-sweet as an example. The developer not only transplanted Apple’s open source XNU kernel, but also successfully launched the iOS 16 user space of iPhone X, including native system services, SpringBoard desktop and other components.

The public development logs of the project also extensively record the process of AI agents such as Claude and Codex participating in kernel modification, driver adaptation, troubleshooting and task handover.

Developers do not need to be familiar with the kernels, startup processes, hardware drivers, and underlying interfaces of different operating systems at the same time. AI can completely let AI read unfamiliar source codes, generate adaptation codes, and complete a large amount of repeated compilation, analysis, and debugging work.

The same goes for game reversal. The unofficial recompilation project of “The Legend of Zelda: Dreamweaver Island” allows the game code originally for Switch to run on Windows, Mac and other platforms. nxnative is trying to take advantage of the fact that Switch and Apple chips both belong to the ARM64 architecture to establish a running environment for Switch games that leads to Mac and iPhone.

As of November 5, the nxnative project has listed more than 120 games that have undergone varying degrees of testing, including Switch exclusive games such as “The Legend of Zelda: Tears of the Kingdom” and “Super Mario: Odyssey”.

Of course, the explosion of these AI reverse engineering projects is, on the one hand, the most powerful proof of the rapid advancement of AI Coding capabilities.But on the other hand, it is also fair to say that AI does not create all technologies from scratch.

Regardless of Apple’s open XNU kernel source code, Android, Linux, Wine, QEMU, and various disassembly and recompilation tools, they have all experienced development for more than ten years or even longer. The role of AI is, to a large extent, to enable individual developers to understand, integrate and utilize these existing results more efficiently.

Coupled with the open source relay on GitHub, the spread of Bilibili and social platforms, if a project achieves a breakthrough, other developers may soon follow.

But being technically easier to do is not enough to explain why people are more willing to do it.

A realistic background is that equipment is becoming more and more expensive and durable. In the past year, the price increase of memory chips has pushed up the cost of mobile phones and PCs. At the same time, flagship mobile phones and graphics cards from a few years ago can still meet many new needs, but the need to replace them is decreasing.

The computing power unlocking of RTX graphics cards is particularly illustrative. In the past, players may not have cared about the throughput limitations of some GeForce tensor calculation instructions. However, after the rise of local large models, AI mapping, and LoRA training, old graphics cards like the RTX 3090 can even achieve more than 60% performance improvement in some specific loads with 24GB of video memory. It is not difficult to understand why they have become popular.

Cracking the limitations of computing performance will naturally bring intuitive value, and it will also give today’s computer development a different practical significance than in the past. Since the hardware can still be used, why can’t it be transformed through software to make it more valuable?

In a sense, this is also an interesting contrast that AI brings to the consumer electronics industry.. On the one hand, AI infrastructure construction is driving up the cost of hardware such as storage, making it more expensive for consumers to purchase new equipment; on the other hand, AI Coding is reducing development costs, making it easier for users to tap the potential of existing equipment.

If the manufacturer wants to stop it, how much can it do?

But for consumer electronics manufacturers, there are obviously some concerns about this wave of AI reverse engineering. On one side are security, compatibility and after-sales responsibilities, on the other side are the software ecosystem, product positioning and even business interests that were originally controlled by the manufacturers.

First of all, we must admit that manufacturers’ control of hardware and software is not completely unreasonable. Modern mobile phones, computers and other personal devices involve a large number of sensitive functions such as payment, privacy, and identity authentication. Unverified system modifications may indeed destroy device security and even cause unpredictable losses to ordinary users.

Game reverse engineering and cross-platform porting also involve more complex intellectual property issues. Nintendo has used litigation and other means to combat Switch emulators and related infringements in the past.

From a technical point of view, manufacturers are not helpless. Locking the bootloader and verifying the system signature can prevent ordinary users from randomly entering other systems. Even if the device is modified, online services such as payments and accounts can control access through integrity verification.

But that doesn’t stop all trolling.Geek Bay Yunfei said in early October that with the help of AI and vulnerability research, all brands of Android retail machines will have the opportunity to obtain Root without unlocking BL.

What’s more, for Mac to run projects such as Android applications and cross-platform recompilation of games, it may not be necessary to modify the system startup chain. Manufacturers can restrict permissions, distribution and services, but it is difficult to treat all compatibility developments as security vulnerabilities. Excessive blocking may also harm the development ecology of the platform itself.

On the other hand, the ideas of AI manufacturers will also have an impact. At the end of October, Anthropic technician Addy Osmani confirmed that the team added relevant instructions to Claude Code’s Opus 5.5 configuration in order to prevent users from using Claude to copy other people’s specific games.

Some users also reported that Claude began to refuse to execute when handling general reverse engineering tasks, and even affected some normal software development and debugging work.

However, questions remain about how effective such restrictions will be.

On the one hand, reverse engineering itself does not equate to piracy. Researching existing software for compatibility involves different legal issues than copying and distributing game content without authorization. On the other hand, Claude is not the only model available to developers. Even if Anthropic restricts certain development behaviors, other AI models, open source tools, and technical exchanges between developers may still promote the continued development of related projects.

The question is, where should these limits be drawn? It is of course necessary to protect user security and respect intellectual property rights, but blindly blocking modification channels may also push users who want to research their own devices into more unsafe exploits.

Especially on highly versatile computing platforms like PCs, users have a long-standing need to install software from other sources and run virtual machines or compatibility layers. If manufacturers continue to tighten restrictions in order to block a few third-party projects, it may also damage the appeal of the platform itself.

And when AI is continuously reducing the development costs of equipment modification and software compatibility, is it still necessary for manufacturers to keep all unofficial modifications out of the system?

Here is a case worthy of reference, which is Fairphone.

Unlike most mobile phone manufacturers, Fairphone clearly supports users to unlock the bootloader and install other operating systems. It will also remind users of related risks and set boundaries for technical support for third-party systems. Of course, this approach may not be suitable for all brands, but it at least proves that manufacturers can ensure default security while retaining options for users who are willing to take risks.

For example, the device is in a secure locked state by default, but users are allowed to actively unlock it through explicit confirmation, clearing original data, etc. For old devices that have ceased official support, you can also consider providing a more convenient unlocking mechanism and necessary technical documentation to help the third-party community extend the software life of the device.

Such changes may, in the short term, conflict with the business model of some manufacturers that rely on the software and hardware ecosystem to promote user replacements and upgrades. But looking at it from another perspective, when the life cycle of hardware is getting longer and longer, users are paying more and more attention to the long-term value of the equipment. Open and sustainable use may also become new brand competitiveness.

The EU smartphone eco-design rules, which will be implemented in June 2025, have already put forward longer-term requirements for device system updates, repair parts and related software support. Although these rules do not require manufacturers to allow users to flash their phones at will, they themselves reflect a direction:

The service life of consumer electronic devices should not be easily determined by software support cycles.

寫在最後

AI Coding has significantly lowered the threshold for complex software engineering, and the open source technology accumulated over a long period of time has more opportunities to be utilized. At the same time, the rising cost of new equipment, extended hardware life cycles, and new demands brought about by AI have also given users more and more motivation to tap the potential of existing equipment.

However, this wave of gaming that has become lively because of AI Coding is still a technology carnival for a few geeks and developers.

Redmi runs iOS, OnePlus runs Hongmeng, and Apple natively runs Switch games… These projects are still far away from normal use by consumers. There are also clear hardware and application scenario restrictions for unlocking the computing power of RTX graphics cards.

As for the cross-platform porting of Switch games, although new progress has been made, there are still many problems before it can achieve a mature and stable user experience and resolve corresponding copyright disputes.