Adobe’s Real Threat Isn’t AI: It’s Losing the Right to Be Irreplaceable

The most important thing about ArtCraft is not whether it can replace Photoshop tomorrow. It is that one developer has made “you can’t build that” sound like a much weaker business argument.

In early October, Brandon Thomas announced seven open-source applications intended to recreate the tools and workflows of Adobe Photoshop, Illustrator, Premiere, Lightroom, After Effects, InDesign, and Acrobat Pro. According to Ars Technica’s report, Thomas used Anthropic’s Claude Opus 5.5, wrote the software in Rust, and made WebAssembly versions available for browsers. The project is distributed under MIT and Apache licenses, although it remains a “super early alpha” with substantial shortcomings.

That last detail matters. ArtCraft is not evidence that Adobe has been technologically defeated. It is evidence that the walls around familiar software are becoming easier to climb—and that changes the negotiation between users and incumbents.

The breakthrough is bargaining power, not perfect parity

The headline claim is seductive: closed applications can now be reconstructed with AI assistance, perhaps by a single developer rather than a large company. Thomas initially said the project would reach 100% feature parity within a month, then softened that forecast, saying 99% parity would take months rather than years. Infobae reported that users were receiving free software under MIT and Apache licenses, but the project’s early state remains central to understanding what has actually happened.

A feature checklist is the wrong measuring stick anyway. Professional creative software is not merely a collection of buttons. It is a dense ecosystem of file formats, shortcuts, plug-ins, bug fixes, color behavior, collaboration habits, documentation, support, and accumulated muscle memory. Recreating the visible surface is much easier than earning trust in a production pipeline.

But incumbents do not need to lose every feature for the threat to be real. They only need to lose their assumption that users have nowhere else to go. A rough alternative can still serve students, hobbyists, independent creators, and organizations that value ownership over polish. More importantly, it can become a platform for contributors to improve rather than a product whose roadmap belongs to one vendor.

This is why the comparison with Adobe is politically and economically sharper than the usual “AI replaces developers” panic. AI is not independently delivering a mature creative suite. It is amplifying a developer who understands the target, makes architectural decisions, evaluates failures, and coordinates iteration. The scarce resource is shifting from typing code to specifying, testing, maintaining, and earning confidence in software.

Closed software’s design problem is the subscription relationship

Thomas says cancellation fees helped motivate the project. That is not a minor anecdote; it is the design critique at the center of the story. When a tool becomes indispensable, the vendor can treat switching costs as a revenue strategy. The user is no longer simply paying for capabilities, but for continued permission to remain inside a workflow.

The University of Chile’s account places ArtCraft alongside wider criticism of subscription practices and notes that Adobe agreed in March 2026 to pay $150 million to the US Department of Justice to resolve a case involving allegedly deceptive practices and hidden early-cancellation charges. Those are concrete reasons users might welcome an alternative, but they also reveal why “enshittification” is too easy an explanation. The problem is not that subscriptions are inherently illegitimate. It is that a subscription can turn dependence into leverage while offering little durable ownership in return.

Open-source reconstruction attacks that leverage from the opposite direction. It says the interface users know is not necessarily the same thing as the company they must rent it from. That distinction is valuable even if ArtCraft never becomes a reliable professional replacement.

Familiarity is powerful—and legally dangerous

There is an important technical distinction between copying code and reproducing behavior. Clean-room reimplementation means studying what a program does, documenting its behavior, and writing an independent implementation. Reverse engineering can be lawful in some contexts, but the boundaries are not universal. The University of Chile source notes that US and European rules protect some forms of interoperability-related investigation while leaving contracts, licenses, and other limits in play.

The legal risk is especially obvious when a clone reproduces not only workflows but also the visual identity of the original. Ars Technica warned that ArtCraft could face problems if its “trade dress”—the recognizable appearance and interface of Adobe’s applications—goes too far. That is a useful warning for the broader open-source movement: openness is not a magic shield against every dispute, and “clean room” is a claim that may be tested rather than a verdict.

This is also where the project’s strategy deserves skepticism. Rebuilding a familiar interface makes adoption easier, but it can make the alternative look dependent on the incumbent’s design language. The strongest long-term competitors will probably borrow concepts users understand while developing their own identities, interaction models, and reasons to stay.

Preservation communities are practicing the same rebellion

The Adobe story belongs in the same conversation as fan-led reconstruction of classic games. The University of Chile reports that AI agents have helped bring three commercial games to 100% exact decompilation, including Super Smash Bros. Melee and both Snowboard Kids titles. Other projects are pursuing ports and reconstructions of games including Halo, Super Mario Galaxy, Conker’s Bad Fur Day, and Fable II.

These efforts are not identical to ArtCraft. Exact decompilation seeks source code that can reproduce an identical binary; static recompilation converts console code into native PC code; clean-room reimplementation starts from observed behavior and builds anew. But culturally they point in the same direction: communities are refusing to accept that software becomes untouchable simply because its original owner has stopped supporting, selling, or documenting it.

That does not make every reconstruction legal, ethical, or sustainable. It does make proprietary control look less permanent. The more tools communities can preserve or recreate, the less credible it becomes for companies to treat compatibility and continuity as favors.

The verdict is not that AI has killed closed software. ArtCraft is too early, and its parity promises remain unproven. The real verdict is more consequential: closed software can no longer rely on scarcity alone to justify its price or lock-in. Adobe and its peers should compete on service, reliability, collaboration, support, and genuine innovation—not on the assumption that users cannot reproduce the workflow.

For readers, the sensible response is neither to abandon professional tools overnight nor to dismiss ArtCraft as a toy. Watch the project, support open alternatives where they meet your needs, and value software that lets you keep control of your work. The winning future may still include proprietary applications. It just should not require treating them as irreplaceable.