Meta made its own AI-generated clickbait news feed Meta 制作了自己的 AI 生成的点击诱饵新闻源

Two Queen Elizabeths appear side by side in an AI-generated image, and that’s not the most disturbing part. The fact that Meta is now systematically manufacturing its own fake news, clickbait, and synthetic content to populate a feed inside its own app—that’s the real story. This isn’t a bug or a rogue experiment. It’s a feature, and it represents the final, cynical surrender of the social media dream.

For two decades, the foundational bargain of social media was this: you give us your attention and data, and in return, we connect you with other people and the world. The feed was a messy, human, sometimes hateful, sometimes beautiful reflection of what people actually wanted to share. Now, Meta is quietly tearing up that contract. With the “For You” section in its standalone Meta AI app, the company isn’t just hosting content; it’s becoming its own content farm, one that runs on electricity and hallucinations instead of freelance writers and click-hungry bloggers. The feed is no longer a mirror of human behavior; it’s a closed-loop system designed to keep you scrolling, with Meta controlling both the show and the audience.

The rationale is brutally simple. Human-generated content is expensive, unpredictable, and comes with PR risks. AI content is cheap, scalable, and controllable. Why host a platform where users might post something offensive or boring when you can generate a limitless supply of content perfectly calibrated to trigger engagement? The "clickbait-style stories" mentioned aren't a flaw in the system; they are the system. They're optimized for the same primal reactions—curiosity, outrage, nostalgia—that have driven traffic since the early days of BuzzFeed listicles. Except now, there's no human at the other end trying to make rent. There's just a model that’s learned the precise formula for "you won't believe what happens next."

This move exposes a deeper, more uncomfortable truth about the trajectory of Big Tech. The "metaverse" bet was a flashy distraction. The real future they're building is a solipsistic one. It’s an ecosystem where the platform doesn’t need its users to create; it only needs them to consume. The original sin of the internet was the democratization of publishing. Meta’s new gospel is the automation of it. They’ve decided the most efficient feed isn’t one that reflects the chaotic internet, but one that supplants it entirely with a predictable, brand-safe stream of synthetic media.

And the quality? It’s telling. The image of two Queen Elizabeths isn’t just a funny error; it’s a perfect symbol. It represents an AI that knows the aesthetics of authority and history but has zero understanding. It can generate the trappings of reality without any grasp of its coherence or meaning. That’s exactly what this feed is: an aesthetic of information without any underlying substance. It’s a hall of mirrors where the reflections are generated by the house itself, and each mirror is subtly, uncannily wrong. We’ve seen AI-generated news articles and images before, but seeing Meta institutionalize them as a core product feature feels like crossing a Rubicon. It’s an official declaration that the truth value of content is irrelevant; only its ability to hold attention matters.

Critics will rightly point out the potential for misinformation. But that misses the bigger, more insidious point. This isn’t about accidentally spreading a false claim. It’s about building a content environment where the very concept of a shared, verifiable reality is undermined because there’s no anchor to anything real. It’s a feed of ghosts, for ghosts. When your “discover” feed is just other AI-generated conversations and images, you’re no longer participating in a social network. You’re interacting with a sophisticated echo chamber that talks back.

This is the endpoint of the engagement-at-all-costs model. When you optimize purely for time-on-app, you inevitably conclude that the ideal user is a passive observer in a world you entirely curate for them. Human messiness is a bug to be patched. The old internet was a bustling, noisy marketplace. Meta is building a pristine, automated showroom where all the products are also generated by the company. It’s cleaner, safer, and utterly soulless.

The removal of the public “Discover” feed—which showed real users’ interactions—is the final clue. That feed was too real. It showed actual human quirks, oddball queries, and the occasional privacy oversight. It was unpredictable. Replacing it with a standard chatbot interface and a bespoke, synthetic content feed is a retreat from the chaos of human connection into the controlled environment of a simulation. You’re not talking to a community anymore; you’re interfacing with a service.

In the end, this isn’t just about one app or one feature. It’s about the direction of a technology that increasingly sees human agency as a inefficiency to be optimized away. Meta isn’t just using AI as a tool; it’s using AI to replace the need for a public square. We’re moving from an internet of people to an internet of services, where your feed is served to you like a meal in a restaurant, with no kitchen tours allowed. And the meal, we now know, is being cooked by the same robots that are serving it. Bon appétit.

当Meta的AI在Facebook时间线上生成两位伊丽莎白女王并存的诡异图片时，这不仅仅是算法失误——这是这家科技巨头在数字内容堕落之路上又一个里程碑式的“杰作”。曾经充斥着外婆转发和阴谋论文章的Facebook，现在终于迎来了它的“进化”：由母公司用AI亲自生产点击诱饵。历史没有终结，历史在循环，而且每次循环都变得更劣质。

Meta AI应用新增的“为你推荐”板块，堪称对信息环境的一次精准投毒。它不再满足于被动接收用户上传的垃圾内容，而是主动、批量、系统性地炮制垃圾。这些AI生成的“故事”，主题往往是耸人听闻的阴谋论、虚假科学或煽情叙事，配图是那种一眼假但足够吸引眼球的AI艺术风格。Meta的逻辑似乎简单粗暴：既然人类用户擅长创造低质量内容，那我们就用AI来实现“降本增效”，把内容农场工业化。用户在时间线上刷到这些AI-generated clickbait时，恐怕会陷入一种认知眩晕：你面对的不是一个人在骗你，而是一个系统在有组织地、规模化地生产欺骗。

这背后是一种令人不安的“堕落螺旋”。几年前，Facebook还因虚假信息泛滥而饱受抨击，不得不增加事实核查。如今，它选择的不是净化信息环境，而是直接加入污染者的行列，甚至想用AI把污染变成一条高产流水线。Meta或许会辩称这只是一个实验功能，但方向已经非常明确：当真实用户生成的内容增长乏力时，就用AI来填充，哪怕填充物是数字合成的“信息垃圾”。这本质上是对用户注意力的掠夺，且不再需要支付人类“写手”的成本。

更讽刺的是，Meta在推动这一功能时，似乎完全无视了其潜在的社会危害。AI生成的虚假王室图片可能只是荒诞的笑话，但当同样的技术被用于制造政治谣言、健康谣言或金融骗局时，其破坏力将呈指数级增长。Meta不是不知道这些风险——它在AI伦理委员会里坐满了专家，但它依然选择将这条危险的路作为产品功能推出。这暴露了其优先级：参与AI军备竞赛、维持应用内用户参与度和时长，远比维护信息生态的健康更重要。

回想Meta AI应用初期，曾因“发现”信息流公开显示用户对话和图像而引发隐私争议，最终被迫下架该板块。现在，它换上了一个标准的聊天界面，看似收敛了，实则将战场转移到了内容生成领域。从暴露用户隐私到生产虚假信息，Meta的AI应用似乎始终在寻找那种能最大化用户互动（无论多有害）的路径。这次，它找到了一个更隐蔽也更系统性的方法：不是让你泄露隐私，而是直接喂给你精心炮制的、迎合你偏见和焦虑的AI内容。

当科技巨头开始用AI制造垃圾来填满自己的平台时，我们看到的不是一个创新的故事，而是一个垄断平台在内容枯竭后的绝望之举。它不再相信或等待真实社区的有机生长，而是选择用合成的数字仿制品来替代。这或许能暂时维持财报上的用户时长数字，但代价是进一步掏空了数字公共空间的真实性。用户面对的将不再是一个由人类分享和交流构成的网络，而是一个由AI生成、旨在捕获点击的幻影剧场。

最可悲的或许是，我们对此几乎无能为力。我们无法阻止Meta在其自己的应用里倾倒AI垃圾，就像我们无法阻止一个餐厅在自己的菜单里添加难吃的菜品。最终的选择权似乎只剩下：继续坐在这个逐渐被合成内容淹没的“餐厅”里，或者离开。但当整个社交网络都朝着这个方向演进时，“离开”本身也成了一个越来越抽象的概念。Meta正在示范的，是一种数字时代的慢性自杀——用AI合成的内容，慢慢毒死使其帝国得以建立的用户信任。