📈 Monday data: Damming the slop floods｜英文原文 + 中文翻译
Exponential View 2026 年 9 月 28 日这一期用六张图表追踪 AI 内容的洪水：网页、音乐、新闻稿的产出已大量由 AI 生成，而听众的注意力远远跟不上——谁来出钱让内容可信？文中给出该期公开正文的英文原文全文与逐段中文翻译。
原文信息
- 英文标题：📈 Monday data: Damming the slop floods。1
- 副标题：AI makes content cheap. Who pays to make it trustworthy?（AI 让内容变便宜，可谁来出钱让它可信？）。1
- 作者：Azeem Azhar、Hannah Petrovic 与 Marija Gavrilov 三人署名（Hannah Petrovic 是 Exponential View 高级研究员，Marija Gavrilov 是董事总经理）。1
- 发布时间：2026 年 9 月 28 日；按官方记录的绝对发布时间，是北京时间 2026 年 9 月 28 日 22:38:32（2026-09-28T14:38:32 GMT），页面最后更新于同日 22:39:53。2
- 本期性质：官方把这一期标为对所有人公开，正文可以一直读到刊尾的五条脚注，页面上没有付费提示。1
- 公开范围：公开部分依次是开场两段、「无限的 AI 内容，需求却是有限的？」「Slop 水坝」「核验预算」三节正文，以及刊尾五条脚注，中间没有会员专享段落。1
- 正文图：六张，依次是 Pew Research 的网页 AI 痕迹曲线、Pangram 的新闻稿 AI 文本比例、Deezer 的 AI 曲目上传与播放对比、Ofcom 的英国成人新闻消费时长、Chartbeat 经 Axios 报道的 Google 引荐流量变化，以及 AI 资金流向（塑造规则与核验产出）。1
- 封面图：本期封面是本频道自制的插图（AI 生成），不取自原文——原文自己的头图就是 Pew Research 那张折线图，属数据图表，不适合作为封面，因此没有采用。1
English original
There’s a flood of AI content coming down the pipeline across all forms of media. Music, images, LinkedIn posts, longform, and book summaries. As AI output grows, the question becomes: what does a future coexistence with AI-generated content look like? 1
The growth of websites made by AI has been noticeable since 2023. Now every third webpage created has some part attributable to AI. 1
Press releases, expected to be a trustworthy source of information, hit an all-time high of nearly a 50% chance of containing AI-generated text in June this year — doubling within a year. AI-generated content and assistance are becoming normalized. 1
Infinite AI content, finite demand?
AI-generated content is not limited to websites and articles, either. The cost of making a music track has dropped so much that, at its peak in June, over half of new music sent to Deezer, a streaming platform, was fully AI-generated. 1
Yet, no one seems to be listening. AI tracks account for just 1-3% of the streams. 1
It’s too early to tell whether this is due to taste or discovery – and we should not infer that humans are not interested in AI music. 1
Audiences seem willing to engage with AI-powered content in the fiction space, and choose to pay. 1
Pocket FM, an Indian audio-drama app, says that AI powers 93% of its content[1]. Its catalog has grown to 2.5 million hours produced a year – from 100,000 hours just two years ago. 1
Revenue had initially stalled in the first six months of overhauling production with AI. The upturn came from feeding listener data back into the process[2]. Ninety-six titles, out of 770,000+, have now made the company more than $1 million each and doubled revenue in a year[3]. 1
Clearly not every title is a hit, but Pocket FM will be a useful test over the next year of where AI input in the creative process creates and retains value. If it does well, it might tell us that, for listeners, it doesn’t matter how content is produced as long as it makes for a good story. 1
Slop dams
UK adults spend some 72 minutes a day consuming news. Just 8 minutes of that are spent accessing publishers directly online. Search, aggregators, AI, social media and video-sharing platforms together account for 37%. 1
Direct exposure to publishers may decline even further. 1
Since Google’s AI Overviews launched in May 2024, small publishers have been the most affected by the piping in the internet’s infrastructure. Even the largest sites lost 22% of their Google referral traffic. 1
Distributors govern entry to platforms and audience discovery. DistroKid handles ~40% of all new music, focused on self-releasing artists, receiving new submissions and pushing them to your favorite music library. At the same time, it hasn’t signed the industry’s anti-fraud standards, while 27 other music and industry bodies have. 1
Universal Music Group (UMG), the largest record label, is now suing DistroKid for allegedly giving platforms and listeners the impression they are getting legitimate music while distributing mass-generated AI content[4]. UMG is not against clearly disclosed AI music. They seem to take issue with what listeners believe is authentic. 1
When Deezer found that in 2025, up to 85% of streams on AI-generated tracks were fraudulent plays, they turned that capability into a business by licensing the AI-detection technology. 1
The verification budget
Music is one of the first, but should not be the last to define what standard is acceptable. Who gets to define what counts as useful AI and whether it should reach audiences? 1
At least $240 million has gone into shaping AI rules and regulations across four disclosed funds in just over a year. AI auditing and certification is a fraction of the funding – and yet it’s the one area that could change how the AI content flood changes the human experience of the internet. 1
The open question is where the responsibility of verification lies. Azeem Azhar recently made the point when interviewed on Bloomberg[5] that “the industry ought to be covering the costs of its own risks, liabilities and damage. That’s what we expect of every industry. And there’s enough money coming in.” Verification should make up part of the budget as a preventative measure. 1
We are entering a world where coexistence with AI-generated content is inevitable, whether in the form of functional content that we tolerate, or in the form of content that we enjoy or something we seek to filter out and avoid. 1
- Pocket FM uses AI across writing assistance, adaptation, narration and production, while humans remain involved in originating ideas and stories. 1
- See Feedback is the moat and Why AI isn’t showing up on your bottomline. 1
- Annual revenue rate. 1
- DistroKid denies the allegations. 1
- Interview starts 26 minutes in. 1
中文翻译
AI 生成的网站从 2023 年起就增长得很明显。如今每新建三个网页，就有一个含有一部分可归因于 AI 的内容。1
新闻稿本被当作可信的信息来源，今年 6 月却创下历史新高：含有 AI 生成文本的概率接近 50%，一年之内翻了一倍。AI 生成的内容和 AI 协助，正在变得习以为常。1
Infinite AI content, finite demand?（无限的 AI 内容，有限的需求？）
AI 生成的内容也不只出现在网站和文章里。做一首曲子的成本降得如此之低，以至于到 6 月的峰值期，发给流媒体平台 Deezer 的新音乐里，超过一半完全是 AI 生成的。1
可似乎没人在听。AI 曲目只占播放量的 1—3%。1
现在判断这到底是因为口味，还是因为曝光不够，还为时过早——我们也不该就此推断人类对 AI 音乐不感兴趣。1
在虚构内容这一块，听众似乎愿意消费 AI 驱动的内容，也愿意掏钱。1
印度的音频剧应用 Pocket FM 说，它 93% 的内容由 AI 驱动[1]。它的内容库已经涨到每年产出 250 万小时——两年前还只有 10 万小时。1
显然不是每个标题都会火，但未来一年里，Pocket FM 会是一个有用的测试：AI 介入创意过程，到底在哪些环节创造价值、留下价值。如果它做得好，也许说明对听众来说，内容怎么生产并不重要，只要故事好看就行。1
Slop dams（slop 水坝）
英国成年人每天大约花 72 分钟看新闻。其中只有 8 分钟是直接访问出版商的网站。搜索、聚合器、AI、社交媒体和视频分享平台加起来，占了 37%。1
出版商拿到的直接曝光，可能还会进一步下降。1
自 Google 的 AI Overviews（AI 概览）在 2024 年 5 月上线以来，小型出版商受互联网这套「管道」的影响最大。即便是最大的网站，也损失了 22% 来自 Google 的引荐流量。1
分发方掌握着进入平台、被受众发现的入口。DistroKid 经手所有新音乐中约 40%，主要面向自行发行的艺人：它接收新提交的作品，再推送到你常用的音乐库里。与此同时，它没有签署行业的反欺诈标准，而另外 27 家音乐机构和行业机构都签了。1
最大的唱片公司环球音乐集团（Universal Music Group，UMG）如今正在起诉 DistroKid，指控它在分发大批 AI 生成内容的同时，让平台和听众以为拿到的是正版音乐[4]。UMG 并不反对明确披露的 AI 音乐。他们介意的似乎是听众以为自己听到的是真人作品。1
Deezer 发现，2025 年 AI 生成曲目上多达 85% 的播放属于欺诈性播放，于是把这项识别能力变成了一门生意：授权这项 AI 检测技术。1
The verification budget（核验预算）
音乐是第一批界定「什么标准可以接受」的领域，但不该是最后一个。谁来界定什么算有用的 AI，以及它是否应该触达受众？1
一年多的时间里，至少有 2.4 亿美元投入塑造 AI 的规则与监管，分布在四支已披露的基金上。AI 审计与认证只占其中一小部分——可偏偏是这一块，可能改变 AI 内容洪水如何改变人类对互联网的体验。1
有一个问题还没有答案：核验的责任该落在谁身上。Azeem Azhar 最近接受彭博社采访时[5]指出：「这个行业应该自己承担自身风险、责任和损害的成本。我们对每个行业都是这么要求的。而且进来的钱足够多。」核验应当作为一项预防措施，纳入预算。1
我们正在进入一个与 AI 生成内容共存不可避免的世界，无论它以我们容忍的功能性内容出现，还是以我们乐于消费的内容出现，抑或是我们要过滤掉、要避开的东西。1
- Pocket FM 在写作辅助、改编、旁白和制作环节都用 AI，而构思创意和故事仍有人参与。1
- 参见 反馈即护城河 与 为什么 AI 没出现在你的利润表上。1
- 指年化收入。1
- DistroKid 否认这些指控。1
- 采访从第 26 分钟开始。1
本文目前公开到这里
上面是这一期官方页面上的全部正文。本期被官方标为对所有人公开，从开场一段到刊尾的第五条脚注都能直接读到，中间没有会员专享段落。1
需要说清楚正文自身的边界：这篇文章以五条脚注收尾，脚注是刊尾的补充说明，不属于正文的延续；文中引用的第三方数据与报道（Pew Research Center、Pangram、Deezer、Ofcom、Chartbeat 经 Axios、以及 AI 资金流向引用的 CNBC、纽约时报、Axios、TechCrunch）都各自指向原文给出的链接，可以逐条核对。1
References
- 1Exponential View 官方详情页
exponentialview.co
- 2Exponential View 官方 RSS
exponentialview.co
- 3Pew Research Center 的分析（原图数据源）
pewresearch.org
- 4Pangram（原图数据源）
pangram.com
- 5Deezer 官方新闻室（原图数据源）
newsroom-deezer.com
- 6Ofcom 报告（原图数据源）
ofcom.org.uk
- 7Axios 报道的 Chartbeat 数据（原图数据源）
axios.com
This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.