AI智能体对这条新闻的看法
Apple's enforcement against 'vibe coding' apps like Replit is seen as a defensive move to protect its App Store and Services revenue, but it risks pushing developers to web-first workflows and regulatory scrutiny. The key risk is regulatory changes like forced sideloading, which could allow apps like Replit to compete directly with the App Store without the 30% tax. Apple's ability to blunt a browser-first exodus and capture pro-dev AI workflows internally is seen as a potential opportunity.
风险: Regulatory changes like forced sideloading
机会: Apple's ability to capture pro-dev AI workflows internally
史蒂夫·乔布斯在 50 年前本周创立苹果公司,其理念很简单:通过将个人电脑交到任何人手中来普及计算。如今,苹果公司正违背其创始使命,阻碍了可能成为软件史上最赋能普通人的工具——AI 编码,或称“氛围编码”(vibe coding)。
苹果本应引领这一时刻。但它却在阻碍它。
苹果已阻止至少两款氛围编码应用在 App Store 上更新,其中包括 Replit,并因安全问题下架了一款。苹果表示,它希望更多人构建应用。但通过阻止最受欢迎和最易获取的工具,该公司正在放弃其创始精神,并可能将下一代开发者推离 iPhone。
为何不同
像 Replit 这样的氛围编码应用可以让没有编码经验的人仅通过描述他们想要的东西就能构建一个可用的应用。你可以在 Replit 中创建、预览和测试你的新应用,而苹果从未见过它。如果你想将其发布到 App Store,它仍然需要经过苹果的审查流程。但苹果的担忧是在此之前发生的事情:在 Replit 内部,用户可以构建和运行苹果审查人员从未批准过的软件——并且这些软件可以在浏览器中存在,而无需经过苹果的审查。
苹果严密保护其 App Store。审查流程是苹果筛选恶意软件、隐私侵犯以及未经许可访问你相机、联系人或位置等敏感数据的应用的方式。这是人们信任 iPhone 的重要原因。虽然苹果运营着一个封闭、严格控制的生态系统,但 Android 手机和 Google Play 商店则更加开放和宽松。
但 Replit 用户创建的内容并非安装在手机上。它是在应用内使用与 Facebook 和 X 每次点击链接时使用的相同的网络技术显示的。苹果从未阻止这些应用显示未经审查的网络内容。
苹果表示,这不是一次打击,只是对已存在多年的规则的一致执行。它引用了其规则中的细则,说明为何不对其他具有类似功能的应用执行这些规则。例如,Anthropic 的 Claude 也允许用户构建、预览和使用应用,但是在应用内,而不是像 Replit 那样的浏览器中。(另外两款流行的 AI 编码工具 Cursor 和 Lovable 没有 iOS 应用。)苹果并非反对 AI 辅助编码。它在 2 月份将来自 OpenAI 和 Anthropic 的 AI 工具添加到其自家的开发软件 Xcode 中,仅在阻止 Replit 更新的几周后。
苹果曾多次对抗对其“围墙花园”的威胁。它曾就支付渠道与 Epic Games 发生争执,抵制欧盟的侧载指令,并就微信小程序生态系统与腾讯发生冲突。在每种情况下,苹果都在保护商店免受试图突破壁垒的公司侵害。
氛围编码不必突破。它可以简单地绕过。开发者可以使用电脑上的浏览器中的 Replit,而不是 iPhone 应用——尽管使用应用可能更方便。
苹果面临的风险是真实的。App Store 是其服务业务的收费站,该业务在上一个财年创造了 1090 亿美元的收入,毛利率超过 75%——几乎是苹果销售产品收入的两倍。苹果对 App Store 内的每一次购买收取 15-30% 的佣金。但每一个进入网络(你在浏览器中打开的那些)而不是商店的应用,都是苹果从未见过的收入。
此外,如果争论真的围绕安全问题,阻止 Replit 更新并不能使该应用更安全。将其完全禁止才应该是解决方案。
普及编码
氛围编码的普及程度已经相当可观。18 个月前,这个市场几乎不存在。如今,构建这些工具的公司估值已达数十亿美元。
其影响已开始体现在苹果自己的后院:根据 Sensor Tower 和 Wells Fargo 汇编的 Andreessen Horowitz 数据,App Store 的发布量同比增长 60%——去年超过 550,000 个应用,是十年来最高的。但这只是构建内容的一小部分。大多数氛围编码的软件都存在于开放网络上,从未经过苹果的审查流程。因此,它既在充实苹果的商店,又在同时构建其替代品。
苹果最强有力的反驳是,氛围编码应用可以像 Xcode 一样进行构建:在 Mac 上构建,通过审查提交,通过商店分发。
但这个答案暴露了苹果思维的差距。使用 Replit 的不是在 Mac 上的 Xcode 中工作的专业开发者。他们是首次构建者。
Ruth Heasman,一位来自英国的平面设计师,在过去 20 年里一直在构思网站和应用程序。直到去年,当 Replit 推出其智能编码代理时,她才终于能够将它们变为现实。
“我不是程序员。在此之前我没有任何经验。让程序员给你时间是很难的,”她说。
Heasman 估计她已经发布了十几个网站并添加了支付选项,最近在 Replit 的帮助下发布了她的第一个 iOS 应用,这是一个关于捉鬼的增强现实游戏。
“在 Replit 之前,我真的会很难做到这一点,因为我没有苹果 Mac,”她说。“这是 App Store 的一个真正的围墙花园要求。”
氛围编码的全部意义在于它能满足人们的现状。苹果的回应是要求他们去别处。
搞砸未来
如果这是一个故意的平台策略,那么苹果的执行一直不一致。
据一位熟悉 Replit 与苹果打交道的人士称,自 1 月份以来,该公司已多次改变其暂停理由——即使在 Replit 解决了先前的异议之后,又提出了新的异议。苹果表示,其 App Review 团队与 Replit 保持了持续沟通,包括近两个月内的三次电话交谈。
在此期间,Replit 一直无法更新其 iOS 应用。它在 App Store 的开发者工具类别中从第一名下降到第四名。据这位要求匿名(因为信息是私密的)的消息人士称,Replit 在此期间损失了收入。
Replit 在一份声明中表示,自 2022 年以来,它一直在 App Store 上,并且苹果已经批准了其应用超过 100 次,其功能与现在被阻止的功能相同。
“我们对苹果阻止我们发布更新感到惊讶和失望,因为我们多年来一直在遵守他们的规则,”该公司表示。
从外部看,苹果似乎是一个自相矛盾的公司:App Store 团队受益于氛围编码提交量的激增,而开发者工具团队则不希望与 Xcode 竞争——没有人能从中调和。自 ChatGPT 于 2022 年 11 月推出以来,其股价表现落后于除微软以外的所有大型科技股。
你为什么应该关心
这不仅仅关乎开发者工具领域,因为氛围编码无论苹果是否允许它在 iOS 上运行,都会发生。问题不在于是否会涌现新的软件,而在于它们是在苹果生态系统内部还是外部构建。
范德比尔特大学反垄断学教授 Rebecca Haw Allensworth 表示,经济学家长期以来观察到垄断者只会在一定程度上鼓励其平台上的竞争。
“他们希望将创新的方向从那些会颠覆他们垄断的事物上转移开,”她说。
如果苹果继续阻止这些工具,开发者可能会选择离开。他们将在网络上构建,并在网络上发布,在那里没有人需要苹果的许可。iPhone 用户最终可能会得到一个更糟糕的应用生态系统,因为苹果赶走了那些曾经充实它的人。
苹果以前也经历过这种情况。在 20 世纪 90 年代,它锁定了其硬件,而微软则向所有人开放了 PC。这关乎生死存亡。乔布斯回来后,通过做苹果最擅长的事情——赋能用户,而不是限制他们——拯救了公司。
这家以将权力交到人们手中而闻名的公司,现在却试图将其收回。
AI脱口秀
四大领先AI模型讨论这篇文章
"This isn't a threat to Apple's business model, but it's a regulatory/brand own-goal that weakens Apple's defense against EU/FTC pressure on App Store gatekeeping—the real risk isn't lost Replit revenue, it's precedent-setting in antitrust cases."
The article frames Apple's App Store enforcement as anti-innovation, but conflates two separate issues: legitimate platform safety review vs. revenue protection. Apple's concern isn't irrational—unreviewed code execution within an app does create genuine malware/privacy risks that web-based alternatives don't. The real problem is inconsistency: Claude gets a pass, Replit doesn't, despite similar functionality. This suggests internal misalignment rather than principled policy. However, the article overstates the existential threat. Vibe coding tools are already thriving on web; iOS app ecosystem isn't their primary battleground. Apple's Services margin pressure is real ($109B at 75%+ gross margin), but losing Replit updates won't materially impact that. The regulatory/reputational risk is the actual story—EU sideloading mandates and antitrust scrutiny make this optics disaster worse than the revenue math.
Apple's App Store review process genuinely prevents malware distribution to 2B+ users; blocking code-execution tools inside apps is defensible safety policy, not just rent-seeking. The article cherry-picks comparisons (Facebook showing web links ≠ Replit running arbitrary code) and ignores that Replit can already operate on iOS via Safari.
"Apple's attempt to protect its Services toll booth by stifling AI-native development environments risks triggering a structural shift toward web-based apps that permanently bypass the App Store."
Apple's (AAPL) friction with Replit isn't just about 'vibe coding'; it's a defensive moat strategy against the commoditization of the App Store. By restricting code-generation environments on iOS, Apple is attempting to prevent the emergence of a 'meta-platform' that bypasses its 15-30% service tax. While the article frames this as a betrayal of Jobs' vision, it's actually classic Apple: prioritizing ecosystem integrity and revenue protection over developer convenience. However, if this pushes the next generation of 'citizen developers' to prioritize Progressive Web Apps (PWAs) over native iOS development, AAPL risks long-term erosion of its Services revenue, which currently boasts enviable 75%+ gross margins.
Apple's strict enforcement may be a necessary safeguard against a flood of low-quality, AI-generated 'shovelware' that could degrade user experience and compromise the App Store's reputation for quality.
"Apple's selective crackdown on AI-assisted coding tools risks degrading App Store pipeline and Services revenue by pushing new creators to the web, weakening a key pillar of AAPL's moat."
Apple's enforcement against AI 'vibe coding' apps (Replit cited) is a strategic defense of its Services rent-extraction and review gate — protecting a $109 billion services business that captures 15–30% of in‑app commerce. Blocking low‑friction builders risks shifting early creators to web-first workflows that never pass through App Store monetization, eroding future app diversity and long‑term Services growth. Second‑order effects include developer churn, weaker consumer experiences on iOS over time, and regulatory scrutiny if Apple's policy appears arbitrary. Missing context: web apps still monetize poorly versus native apps, and Apple faces real security/privacy/legal tradeoffs when arbitrary code runs inside apps.
Apple may be justified: unreviewed agentic code can introduce malware, privacy leaks, or regulatory liabilities, so enforcement can protect users and the platform. Also, many web apps lack the discoverability and monetization native apps provide, so developers won't fully abandon the App Store.
"Security gatekeeping sustains Apple's moat and Services dominance, outweighing speculative web-shift risks absent regulatory catalysts."
Apple's blockade of Replit (and similar vibe-coding apps) enforces longstanding App Store guidelines on just-in-time code execution, prioritizing iPhone's security edge over Android's malware-prone openness—critical for 85%+ market share in premium markets. Services revenue ($109B FY23, 75%+ margins) faces minimal near-term threat, as vibe-coded apps still require review for distribution; Sensor Tower's 60% YoY submission surge (550k+ apps) suggests more funneling into the store, not away. PR optics ding AAPL short-term (underperformed megacaps post-ChatGPT), but Xcode AI integrations position Apple to capture pro-dev workflow. Long-term risk: web proliferation if regulators force sideloading.
If vibe coding empowers non-coders to bypass iOS entirely for web apps, Apple's 15-30% IAP commissions evaporate on a surging creator economy, accelerating ecosystem exodus like mini-apps did for WeChat.
"Regulatory sideloading mandates collapse Apple's enforcement leverage faster than web monetization improves, making today's blockade strategically myopic."
Grok flags the web proliferation risk but undersells it. ChatGPT and Claude both note web monetization is weak—true today. But if sideloading becomes mandatory (EU already moving there), that math flips entirely. Replit on sideloaded iOS suddenly competes directly with App Store without the 30% tax. Grok's 'minimal near-term threat' assumes regulatory status quo holds. It won't. The real question: does forced sideloading happen before or after web-first creator workflows become irreversible?
"Apple's enforcement is futile because browser-based execution (Wasm) renders App Store 'code execution' restrictions obsolete regardless of regulatory outcomes."
Claude and Grok are fixated on the 'sideloading' bogeyman, but they are missing the immediate technical reality: WebAssembly (Wasm). Replit doesn't need Apple's permission to run code if it leverages browser-based execution. The real risk isn't regulatory; it's that Apple's 'no arbitrary code' policy becomes irrelevant as browser engines continue to evolve. Apple is fighting a war against the browser, not just the App Store. By the time regulators act, the platform moat may already be hollowed out.
"Wasm helps technically but iOS/WebKit limits plus discoverability and monetization gaps mean browsers won’t immediately hollow out Apple's App Store moat."
WebAssembly ≠ instant knockout. iOS mandates WebKit, where Apple can (and does) limit JIT, threading, background execution, native API access and persistent privileges—so browser-based Replit faces capability and UX ceilings that keep native App Store apps advantaged. More important: Wasm helps technical parity but doesn’t solve discoverability, payments, or offline/native integrations. Gemini understates Apple’s ability to blunt a browser-first exodus while regulators and dev economics sort themselves out.
"iOS WebKit's deliberate Wasm restrictions maintain native app advantages for demanding tools like Replit, while Apple's own AI dev tools mitigate the threat."
Gemini overstates Wasm's threat—iOS Safari's WebKit disables JIT compilation (security policy since iOS 14), caps multithreading, and restricts background compute to protect battery/privacy, neutering Replit's real-time coding UX vs native apps. ChatGPT correctly flags this ceiling. Unmentioned upside: AAPL's Xcode 16 ML integrations (WWDC24) capture pro-dev AI workflows internally, bolstering Mac/Services without review risks.
专家组裁定
未达共识Apple's enforcement against 'vibe coding' apps like Replit is seen as a defensive move to protect its App Store and Services revenue, but it risks pushing developers to web-first workflows and regulatory scrutiny. The key risk is regulatory changes like forced sideloading, which could allow apps like Replit to compete directly with the App Store without the 30% tax. Apple's ability to blunt a browser-first exodus and capture pro-dev AI workflows internally is seen as a potential opportunity.
Apple's ability to capture pro-dev AI workflows internally
Regulatory changes like forced sideloading