Back in March, we wrote a post about a significant application of AI to re-write an open source project. That in itself was noteworthy, but the decision to release the re-written code under a different license had even bigger implications for the whole of open source. Using AI in this way meant that any open source code could be relicensed under any license – or, indeed, no license – a point confirmed by Bruce Perens, who wrote the original Open Source Definition. That, in its turn, meant that the use of copyright law by open source projects to prevent precisely this kind of relicensing was effectively pointless. The article focused on the impact of this application of AI on the open source world. But I also made a prediction:
These kind of “AI clean room” implementations could be used to make new versions of any proprietary software.
A little sooner than I expected, that is precisely what has happened. The developer Brandon Thomas announced last month on Reddit:
I’ve created Open Source 100% Rust clean-room replacements for 7 of Adobe’s apps: Photoshop, Illustrator, Premiere, Lightroom, After Effects, InDesign, and Acrobat Pro. Free, no subscription, and you own the source code
Thomas used Anthropic’s Claude Opus 5.5 to create software that is designed to replicate the features of Adobe’s well-known applications. Unsurprisingly, the fidelity is only partial, and people have criticized the current state of the replacements. Thomas explained that:
These are still super early alpha. All MIT/Apache licensed. No telemetry.
I’m going to try to drive these to 100% parity but it’s going to take a minute. And I’ll need lots of help. (Anyone want to join me? I’ll pay stipends and bug bounties.)
Rather boldly, he also added:
I’m going to do Microsoft Office, AutoCAD, and Solidworks too. Probably a lot of stuff to collapse.
And in fact ArtCraft, the company of which Thomas is CEO, now offers WordCraft, GridCraft, DeckCraft and CADCraft, all “Native, open-source apps from the ArtCraft team, built in Rust and free to use.” It is claimed that the apps all run natively on Windows, macOS and GNU/Linux.
All those apps are still under development, and presumably getting better, although it’s not clear what the final results will be – for example, whether there are major gaps in functionality that just can’t be filled. But in many ways, the details don’t matter. What Thomas’s project has shown is that it is possible to use AI to produce apps that replicate at least some, possibly most, of the functionality of leading commercial software packages. People are already working on the next stage: generic AI tools that can be used to explore existing software and then replicate its features automatically, and without looking at the source code. One such tool is REA: Reverse Engineer Anything, which uses AI agents:
See a feature in an app that you want in your own product? Ask your agent to investigate it with REA. It can inspect the app without its source code, explain how the feature works, show the evidence, and build a version for your project.
REA connects your agent to tools for inspecting native binaries, JavaScript and Electron apps, .NET assemblies, and websites. You can also use the same tools from your terminal. Analysis runs locally, and results include the evidence and limitations behind each conclusion.
The big breakthrough here is that AI does all the hard work required for “clean room” reverse engineering. Historically, this was an extremely slow and costly process, which meant that it was only carried out for major projects where the benefits would be correspondingly great. AI means the cost and effort of doing so has dropped by many orders of magnitude, making all kinds of new projects feasible for the first time, including producing clones of major software packages.
These latest developments have huge implications for many industries. As with the open source case discussed earlier this year, they effectively make copyright protection for commercial software moot. There are already the first signs that AI companies are waking up to this issue, and are trying to stamp it out, reported here by Tom’s Hardware:
we had already linked to recent issues raised at Claude Code’s own GitHub repository, comprised of users complaining that the AI assistant was refusing to work on their games or on reverse-engineering tasks. As far as we can ascertain, it appears the game-related blocking is limited to the combination of using Claude Code and the Opus 5.5 model.
…
Regarding the blocking of reverse engineering, we have received direct confirmations. The first comes from our own alumnus Jarred Walton, who was trying to run checks on the software he’s developing. He notes that such work had been performed with no issue until yesterday, October 31, and that switching to Opus 4.8 let him continue his work.
Walton theorizes that Anthropic may have added a new rule to Code specifically for reverse-engineering.
Given the potential impact of reverse engineering not just on the software industry, but everywhere that software is used in products, it is likely all the big AI companies will start trying to ban reverse engineering to head off industry and government criticisms. This will turn into the usual cat and mouse game as people seek and find workarounds, which are then blocked. But all such moves will do is encourage people to use open source and open weight AI systems, for example Chinese ones. The latter are now so powerful that little will be lost in terms of capability, and much gained in terms of freedom.
However, replicating Adobe’s and Microsoft’s functionality and familiar interfaces is bound to face legal challenges. In addition to the numerous court cases that are considering alleged copyright infringement by AI models of words, images and sounds, we can expect to see new lawsuits that claim software copyright is being infringed in various ways by this new wave of AI reverse engineering.
Follow me @glynmoody on Mastodon and on Bluesky. Originally posted to Walled Culture.