AI music is flooding streaming services — but who wants it?
AI music is flooding streaming services — but who wants it?
AI 音乐正涌入流媒体平台——但谁在买账？
They won’t ban it. They won’t embrace it either. 它们不会封杀，也不会拥抱。
By Terrence O’Brien | May 3, 2026, 12:00 PM UTC 作者：Terrence O’Brien | 2026年5月3日，世界标准时间中午12:00
Terrence O’Brien is the Verge’s weekend editor. He has over 18 years of experience, including 10 years as managing editor at Engadget. This is The Stepback, a weekly newsletter breaking down one essential story from the tech world. For more on how AI is changing music and the music industry, follow Terrence O’Brien. The Stepback arrives in our subscribers’ inboxes at 8AM ET. Opt in for The Stepback here. Terrence O’Brien 是《The Verge》的周末编辑。他拥有超过 18 年的行业经验，其中包括在 Engadget 担任执行主编的 10 年经历。《The Stepback》是一份每周通讯，旨在深度剖析科技界的一个核心故事。想了解更多关于 AI 如何改变音乐及音乐产业的内容，请关注 Terrence O’Brien。《The Stepback》会在美东时间上午 8 点发送至订阅者的收件箱。点击此处订阅《The Stepback》。
How it started
起源
The use of generative AI in pop music started almost as a gimmick. There was a sense of experimentalism to 2018’s I AM AI by Taryn Southern and 2019’s Proto by Holly Herndon, albums that were created with significant assistance from AI. Others got in on the action too, exploring the outer limits of tools like Google’s Magenta and even training their own models. But things quickly changed with the launch of Suno in December of 2023 and Udio in April of 2024. 生成式 AI 在流行音乐中的应用最初几乎像是一种噱头。Taryn Southern 2018 年的专辑《I AM AI》和 Holly Herndon 2019 年的《Proto》都带有实验性质，这些专辑在创作过程中得到了 AI 的大量辅助。其他人也纷纷加入，探索 Google Magenta 等工具的极限，甚至开始训练自己的模型。但随着 2023 年 12 月 Suno 和 2024 年 4 月 Udio 的发布，情况迅速发生了变化。
Suno and Udio allow users to quickly create entire compositions with a simple text prompt. AI-generated music was no longer the realm of technical experts and fringe experimenters, it was now accessible to anyone with an internet connection. This led to an influx of machine-made music hitting streaming platforms. Suno 和 Udio 允许用户通过简单的文本提示快速创作出完整的音乐作品。AI 生成音乐不再是技术专家和边缘实验者的专属领域，现在任何能上网的人都可以使用。这导致大量机器创作的音乐涌入流媒体平台。
In September of 2025, Deezer said that 28 percent of music uploaded was fully AI-generated. By the end of the year, that had grown to over 50,000 tracks per day, accounting for 34 percent of uploads. Both users and artists have expressed frustration, demanding streaming platforms do something to combat the growing problem that is watering down playlists and siphoning millions in royalties away from legitimate artists. Udio did not reply to a request for comment. 2025 年 9 月，Deezer 表示其上传的音乐中有 28% 是完全由 AI 生成的。到年底，这一数字增长到每天超过 5 万首曲目，占上传总量的 34%。用户和艺术家都表达了不满，要求流媒体平台采取行动，解决这一日益严重的问题——它正在稀释播放列表，并从合法艺术家手中窃取数百万美元的版税。Udio 未回复置评请求。
How it’s going
现状
Things have only gotten worse at Deezer, where daily uploads of AI-generated content have grown to 75,000, and are threatening to overtake actual human-made music. And Spotify removed over 75 million spam tracks in just 12 months. Deezer 的情况变得更糟，AI 生成内容的日上传量已增至 7.5 万首，正威胁要超过人类创作的音乐。而 Spotify 在短短 12 个月内就删除了超过 7500 万首垃圾曲目。
Deezer was the first major streaming platform to implement a system that detects and labels AI-generated content. The service also prevents its algorithm from recommending it and has demonetized 85 percent of the streams. In a recent press release, Deezer CEO Alexis Lanternier said that, “AI-generated music is now far from a marginal phenomenon and as daily deliveries keep increasing, we hope the whole music ecosystem will join us in taking action to help safeguard artist’s rights and promote transparency for fans.” Deezer 是第一个实施 AI 生成内容检测和标记系统的主要流媒体平台。该服务还阻止其算法推荐此类内容，并对 85% 的此类流媒体播放进行了去货币化处理。在最近的一份新闻稿中，Deezer 首席执行官 Alexis Lanternier 表示：“AI 生成音乐早已不再是边缘现象，随着每日上传量的持续增加，我们希望整个音乐生态系统能与我们一起采取行动，帮助维护艺术家的权利，并为粉丝提高透明度。”
Qobuz was next to implement a detection system. It also published an AI charter, promising that it would never use AI for its editorial or curation content. While the company stopped short of banning AI-generated content, it did lean into the discontent, saying, “The heart of Qobuz is and will remain human.” Qobuz 是下一个实施检测系统的平台。它还发布了一份 AI 章程，承诺绝不将 AI 用于其编辑或策展内容。虽然该公司没有完全封杀 AI 生成内容，但它顺应了不满情绪，表示：“Qobuz 的核心是，且永远是人类。”
Apple soon followed. Though its labeling system has an obvious flaw — it relies on self-reporting. Apple Music “requires” labels and creators to voluntarily add Transparency Tags to their metadata. When asked how it was enforcing requirements, or what penalties, if any, there were for failing to label AI-generated content, Apple declined to comment and pointed me to an industry newsletter from early March that says it “defers to content providers to determine what qualifies as AI content.” 苹果紧随其后。尽管其标记系统有一个明显的缺陷——它依赖于自我申报。Apple Music “要求”唱片公司和创作者自愿在元数据中添加透明度标签。当被问及如何执行这些要求，或者未能标记 AI 生成内容会有什么惩罚（如果有的话）时，苹果拒绝置评，并向我指出了一份 3 月初的行业通讯，其中称苹果“听从内容提供商的判断，由其决定何为 AI 内容”。
Spotify also opted for a voluntary system. It recently launched AI credits, which identify tracks made using generative AI. It’s working with the standards group DDEX to create an industry standard for labeling AI content. It goes beyond blanket labeling, allowing artists to specify whether AI was used to create the lyrics, vocals, or backing music. Initial glimpses of those efforts began rolling out in mid-April, with DistroKid as its first partner. Spotify 也选择了自愿系统。它最近推出了 AI 信用额度（AI credits），用于识别使用生成式 AI 创作的曲目。它正在与标准组织 DDEX 合作，制定 AI 内容标记的行业标准。这不仅仅是简单的统一标记，还允许艺术家具体说明 AI 是否被用于创作歌词、人声或背景音乐。这些工作的初步成果已于 4 月中旬开始推出，DistroKid 是其首个合作伙伴。
While DDEX counts most of the industry’s heavyweights as members — including Amazon, Google, Meta, Apple, Songtradr (home of Bandcamp), Pandora, BMI, UMG, Sony Music Entertainment, and Warner Music Group — not everyone is necessarily on board with Spotify’s standard yet. 虽然 DDEX 的成员包括大多数行业巨头——包括亚马逊、Google、Meta、苹果、Songtradr（Bandcamp 的母公司）、Pandora、BMI、环球音乐、索尼音乐娱乐和华纳音乐集团——但并非所有人都一定认同 Spotify 的标准。
Spotify has come under criticism for its handling of AI slop and so-called ghost artists. But recently, it has gone out of its way to talk up its transparency efforts and its increased offensive against spam and impersonation. The company also recently launched a Verified by Spotify badge that is supposed to guarantee there’s a human behind an artist profile. Sam Duboff, global head of Marketing & Policy, Spotify for Artists, told The Verge that it is experimenting with third-party detection tools, but that they still make a “material amount of incorrect assessments.” Spotify 因其对 AI 垃圾内容和所谓“幽灵艺术家”的处理方式而受到批评。但最近，它不遗余力地宣传其透明度工作，并加强了对垃圾内容和冒充行为的打击力度。该公司最近还推出了“Spotify 认证”徽章，旨在保证艺术家资料背后是真实的人。Spotify 艺术家营销与政策全球主管 Sam Duboff 告诉《The Verge》，他们正在试验第三方检测工具，但这些工具仍然会做出“相当数量的错误评估”。
Google also requires that AI-generated content be labeled, whether that’s on YouTube or YouTube Music. While the company won’t publicly detail how its systems to combat AI slop work, it has said that it’s “building on… established systems that have been very successful in combatting spam and clickbait, and reducing the spread of low quality, repetitive content.” It also says that failing to disclose can carry penalties, including the removal of content or suspension from the YouTube Partner program. Google 也要求对 AI 生成的内容进行标记，无论是在 YouTube 还是 YouTube Music 上。虽然该公司不会公开其打击 AI 垃圾内容的系统运作细节，但它表示，它正在“基于……在打击垃圾邮件和点击诱饵方面非常成功的既有系统，并减少低质量、重复内容的传播”。它还表示，未披露信息可能会受到处罚，包括删除内容或暂停 YouTube 合作伙伴计划资格。
In survey after survey, public opinion toward AI music is pretty unfavorable. A study by Deezer and Ipsos showed that 51 percent of respondents think AI will “lead to the creation of more low-quality, generic-sounding music.” 在一项又一项的调查中，公众对 AI 音乐的看法相当负面。Deezer 和益普索（Ipsos）的一项研究显示，51% 的受访者认为 AI 将“导致更多低质量、听起来千篇一律的音乐产生”。
A poll conducted by The Hollywood Reporter and the Frost School of Music found that 66 percent of people never knowingly listen to music generated by AI. And 52 percent said they wouldn’t even want to listen to music from their favorite artist if they knew it was made with help from AI. 《好莱坞报道》和弗罗斯特音乐学院进行的一项民意调查发现，66% 的人从未主动听过 AI 生成的音乐。52% 的人表示，如果知道自己最喜欢的艺术家的音乐是在 AI 的帮助下创作的，他们甚至不想再听了。
Researchers from Singapore also found significant negative bias toward AI-generated content. 来自新加坡的研究人员也发现，人们对 AI 生成的内容存在显著的负面偏见。