A usable Rust image editor may now exist — but the more interesting story is the legal territory the code walks into.
An effort that Phoronix is reporting under the name ArtCraft — with the outlet's URL slug suggesting the individual implementations may carry the names Photocraft and Filmcraft — has produced open-source, Rust-language reimplementations of well-known Adobe creative applications: Photoshop for image editing, Premiere for video editing, and Lightroom for photo management. Phoronix says the projects were built with significant assistance from Claude Code and other AI agents.
That alone is a meaningful data point for anyone tracking the open-source creative-tooling landscape. Photoshop and Premiere are among the most technically demanding applications in consumer software, with decades of accumulated feature work behind them. The conventional route to competing with them — a multi-year, multi-team engineering effort — has long put such applications beyond the reach of hobbyist and community projects. Phoronix's framing suggests that route is being compressed by AI agents capable of generating large volumes of functional code quickly.
What the project actually is
According to Phoronix's report, the projects are implemented in Rust, the systems programming language that has gained strong traction in performance-sensitive open-source work — including the Linux graphics stack — for its memory-safety guarantees. The report characterises them as "open-source, clean-room reimplementation" versions of Adobe's products. The exact project naming remains somewhat unclear from the coverage alone — Phoronix's summary text refers to "ArtCraft," while the article's own URL slug points to Photocraft and Filmcraft — so the project name should be read as provisional pending direct verification.
What is not in question is the core claim as Phoronix presents it: this is the project's own description of its approach, not a legal finding. Nothing has been adjudicated, and no court has reviewed the codebase.
The "clean-room" question
Glossary: A clean-room implementation is the practice of recreating the functionality of an existing product without copying its source code or protected expression — typically by having one team study the original and document its observable behaviour, while a separate, legally separated team writes new code from those specifications alone. Traditionally, the point is to guarantee independence from the original author's creative expression.
Copyright protects expression, not functionality — which is why reusing an application's features has generally been permissible, provided the code was written independently. That principle, in theory, should favour reimplementation.
The difficulty is that AI-assisted development may not produce the same kind of independence a clean-room process is designed to guarantee. If an agent trained on large corpora — which may include proprietary or otherwise licensed material — generates the implementation, the traditional evidentiary basis for a clean-room claim weakens. There is, as far as the reporting indicates, no established legal precedent settling how AI-generated code interacts with clean-room doctrine, or how training provenance should be assessed.
Separately from copyright, Adobe's trademarks, trade dress, and patents attach to those assets regardless of how the code was produced. There is no indication, per Phoronix's account, that Adobe has responded to, endorsed, or challenged the project — speculation about the company's likely intentions at this stage is premature.
What to watch
The practical takeaway is not that the gap between open-source and commercial creative tools has closed; limitations in such projects tend to surface quickly, and the reporting treats these implementations as early-stage rather than complete substitutes. The durable question is whether the codebase survives scrutiny — technical, community, and legal.
For IT professionals in Hong Kong and elsewhere, the case is a live illustration of a broader trend already visible in Linux, security tooling, and developer infrastructure: AI agents are materially changing the economics of building substantial software, and the legal frameworks governing that shift are still catching up.
The open question is no longer whether AI can write a usable application in Rust. It is whether that application can withstand the scrutiny a commercially significant product eventually invites.
一個可用的 Rust 圖像編輯器也許已經存在——但更值得關注的是這段程式碼所走進的法律領域。
據 Phoronix 報道，一個以「ArtCraft」之名發布的計劃——雖然該網站的 URL slug 暗示各項重寫版本可能分別命名為 Photocraft 和 Filmcraft——以 Rust 程式語言重寫了多款廣為人知的 Adobe 創作應用程式：包括用於圖像編輯的 Photoshop、影片編輯的 Premiere，以及相片管理的 Lightroom。Phoronix 表示，各項目在 Claude Code 等 AI agent 的顯著協助下完成。
單是這一點，對任何關注開源創作工具生態的人來說，已是值得重視的指標。Photoshop 和 Premiere 是消費級軟件中技術難度最高的應用程式之一，背後累積了數十年的功能開發。傳統的競爭路線——動員多個團隊、耗時多年的工程努力——長期令此類應用程式超出業餘愛好者和社群項目所能觸及的範圍。Phoronix 的報道框架暗示，這條路線正被 AI agent 壓縮，後者能夠快速生成大量具實用功能的程式碼。
該計劃的真實內容
根據 Phoronix 的報道，各項目以 Rust 實現——這門系統程式語言憑藉其記憶體安全保障，在對性能要求嚴謹的開源工作中（包括 Linux 圖形技術堆疊）取得了強勁發展。報道將它們描述為 Adobe 產品的「開源、乾淨室重寫版本」。單從報道內容來看，項目的確切命名仍不甚明確——Phoronix 的摘要文字使用「ArtCraft」，而文章本身的 URL slug 則指向 Photocraft 和 Filmcraft——因此項目名稱應視為暫定，有待直接核實。
確定無疑的是 Phoronix 所呈現的核心論述：這是該計劃對自身方法的描述，並非法律裁定。目前未有任何司法裁決，也沒有法院審視過該程式碼庫。
「乾淨室」問題
名詞解釋： 乾淨室重寫（clean-room implementation）是指在不複製原有產品的原始碼或受保護表達的情況下，重建其功能的實踐——通常做法是由一組團隊研究原有產品並記錄其可觀察的行為，然後由另一組在法律上分離的團隊，僅根據上述規格撰寫全新程式碼。傳統上，這種做法的目的是確保與原作者的創意表達保持獨立。
版權保護的是表達，而非功能——這就是為何在程式碼獨立撰寫的前提下，重新利用應用程式的功能一般而言是被允許的。這原則在理論上應有利於重寫。
困難在於，AI 輔助開發未必能產生與乾淨室流程所要保證的同一種獨立性。如果是由一個在大型語料庫（其中可能包含專有或其他受授權材料）上訓練的 agent 生成該重寫版本，那麼支持乾淨室主張的傳統證據基礎便會削弱。就報道所顯示的情況而言，目前並無確立的法律先例來界定 AI 生成的程式碼如何與乾淨室原則互動，以及應如何評估訓練資料的來源。
版權以外，Adobe 的商標、商品外觀（trade dress）和專利，不論程式碼是如何生成的，都與該些資產相掛鉤。按 Phoronix 的描述，目前未有跡象顯示 Adobe 已就該計劃作出回應、背書或提出質疑——在此階段猜測該公司的意圖仍言之尚早。
值得關注的事態
實際的啟示並非開源與商用創作工具之間的差距已經收窄；此類項目常見的局限往往很快浮現，而報道亦將這些重寫版本視為早期階段的嘗試，而非完整的替代品。持久的問題在於，該程式碼庫能否經受各種審視——技術上的、社群上的，以及法律上的。
對香港及其他地區的 IT 專業人士而言，此案例生動地體現了一個已在 Linux、安全工具和開發者基建領域顯現的更廣泛趨勢：AI agent 正在實質性地改變開發大型軟件的經濟效益，而規管這轉變的法律框架仍趕不上形勢。
懸而未決的問題，已不再是 AI 能否用 Rust 寫出一個可用的應用程式，而是該應用程式能否經受住一款具商業重要性的產品最終所引來的審視。