---
author: Internet of Bugs
date: '2026-10-08'
guest: ''
layout: post.njk
source: https://www.youtube.com/watch?v=GtBwaxwFpg0
speaker: Internet of Bugs
tags:
- media-ethics
- source-transparency
- deepfake
- disinformation
- brandolinis-law
title: AI时代主流媒体的透明度塌陷：以《纽约时报》与Ezra Klein为例
summary: 软件工程师Carl批评以《纽约时报》及Ezra Klein为代表的传统媒体在AI报道中极度缺乏信源透明度。随着AI深度伪造让虚假信息泛滥，主流媒体长期的懒惰与傲慢不仅破坏了事实核查基础，更让辟谣成本呈数量级上升。
insight: ''
draft: true
series: ''
category: politics-society
area: society-thinking
project: []
people:
- Ezra Klein
companies_orgs:
- The New York Times
products_models: []
media_books: []
status: evergreen
---
### 透明度沦陷：AI时代新闻媒体的懒惰与傲慢
在人工智能技术飞速渗透的时代，媒体所展现出的不负责任令人堪忧。这种不负责任并非仅仅体现在一贯对 AI 炒作的推波助澜上，而是指向一个更为根本且普遍的顽疾：以 **Ezra Klein** 及 **《纽约时报》**（The New York Times）为典型代表的主流媒体，已然傲慢且懒惰到不愿采取最基础、最显而易见的透明度举措。这使得那些试图通过呈现事实另一面来维持行业诚信的独立创作者陷入了极度被动的境地。
通常情况下，过度使用“假新闻”（Fake News）这一叙事极易被掌握权力的政商利益集团利用，用以掩盖真相并逃避问责。但在当下的现实中，AI 技术让制造虚假信息变得极为低门槛且难以察觉，如果社会文明还要维系，就必须要求媒体至少恪守最底线的新闻标准。面对主流媒体对信息来源的漠视，公众理应得到更严肃的对待。作为自 20 世纪 80 年代末便从业的软件专业人士，**Internet of Bugs** 的创办者 Carl 指出，对抗 AI 炒作与媒体懒惰不仅是为了防止互联网演变为更加充满漏洞和信任危机的噩梦，更是一场捍卫事实证据的持久战。
Original English Source
Today, I'm going to talk about the irresponsibility of the media in the age of AI. For once on this channel, though, I'm not going to be complaining about the media promoting AI hype, although they're still doing that too, of course. No, today, I need to talk about a much, much bigger and more pervasive problem. What I'm talking about today is way more basic than the AI hype. It's that the media, exemplified in this video today by Ezra Klein and The New York Times, is too lazy and arrogant to care about taking really obvious and simple steps to be transparent, which in turn is making it much, much harder for those of us trying to keep them honest by giving you the other side of the story.
I really don't like lending credibility to the "Fake News" narrative, because quite often that phrase is used by the powerful government and business interests to try to obscure facts and avoid accountability for their actions. But, in the AI world we find ourselves in - which will almost certainly continue for however long we manage to hold on to civilization - as the new AI tech makes creating fake news incredibly easy to do, and incredibly difficult to detect, it's time we started holding the media to at least a bare minimum standard. And this is long overdue. Decades overdue. The public deserves better. You deserve better. I'm so sick and tired of busting my ass, trying to make it clear to you where the information I use in my videos is coming from when most of the American media just doesn't give a F---.
This is Internet of Bugs. My name is Carl. I've been a software professional since the late 1980s, and I'm trying to do my part to keep AI from making the internet need to be even more bug ridden and untrustworthy nightmare than it already is. You can find more information about me at InternetofBugs.com, and if you want to help support my work trying to push back against the AI hype and media laziness, there's a link to my Patreon below. I truly appreciate all my patrons and all the people who have helped me out in any of various ways. This has been a really uphill battle.
### 伪装成严肃报道的点击诱饵与信源黑洞
触发这次声讨的导火索，是 Ezra Klein 近期在《纽约时报》发表的一篇视听评论。该作品不仅再度充当了 AI 产业的传声筒，鹦鹉学舌般兜售行业宣传与末日论（Doomerism），更暴露了其新闻专业主义的严重缺失。《纽约时报》在 YouTube 频道的主页上宣称自己是“最强大的独立实地调查与深度报道引擎”，并自诩“树立了行业标准”。然而在其涉及科技与 AI 领域的报道中，这种所谓的标准实质上沦为了彻头彻尾的点击诱饵，以及科技专栏作家 Karl Bode 所定义的**“CEO随口一说”式新闻**（"CEO said a thing" journalism: 盲目照搬企业高管言论而缺乏实质求证的新闻范式）。
为了纠正主流媒体制造的偏见与错误叙事，独立研究者必须穿透耸人听闻的标题和封面图，逐一深挖支撑其观点的底层事实并公之于众。然而，传统媒体的不透明使得这种事实核查工作变得异常艰难且耗时：
* **零信源外链**：在时长近 30 分钟的视频短片中，外部素材与引用片段占比高达三分之一（约 9 分钟），但在视频简介栏、文字实录甚至《纽约时报》官网页面中，没有提供任何可追溯的原始链接。
* **官方链接损坏**：视频简介中唯一给出的文字实录链接直接导向 404 错误页面，唯一正常工作的 URL 竟是其播客的图书营销推广链接。
* **敷衍的素材致谢**：片尾仅以三张杂乱无章的滚动截屏列出了十余家组织与机构（包含 Anthropic、Google、OpenAI、多家新闻台、TikTok 创作者与 Twitter/X 用户），完全没有标注具体片段与对应出处的对应关系。
Original English Source
I started making the video you are now watching when I encountered a recent piece by Ezra Klein from the New York Times carrying water for the AI industry and parroting their propaganda and Doomerism again. And it just kind of broke me, which is why instead of calling out the New York Times for shilling for the AI billionaires again, I find myself making a video calling out the New York Times for just being lazy, lousy journalists. And it's not just them, just to be clear. So much more of the media industry is just as bad as the New York Times.
But on their YouTube page, the New York Times called themselves the most powerful engine for independent boots on the ground and deeply reported journalism, and they claim we set the standard. And in some sense, it's probably true that they set the standard. But given how pathetic their standard is, I don't think they ought to be bragging about it. Based on the examples we're going to talk about today, their most powerful engine is, well, at least on the subjects of AI and tech or I work, pretty much just clickbait and what Karl Bode calls "CEO set a thing" journalism.
Some of us in our own way, in our own niches, try to do our best to counteract their misinformation, not that we can ever fix all the damage they cause. But to correct the record, what we have to do is to take the media sensationalist clickbait headlines and thumbnails and articles, and then we have to drill down to get to the underlying verifiable facts that the media narrative is based on, and then point out to you, the public, what those facts actually are. And the media through their laziness and arrogance makes that so much more difficult and time consuming than it ought to be.
So here's a link to the Ezra Klein video essay that I'm using in my primary example today. As always, you'll find all my links to all my references below. Any time today that I'm showing you clips or stills from this video, you'll see the timestamp in seconds up here on the top right corner, and the YouTube plus captions hard coded down here in the bottom left. The video is just short of 30 minutes long, and this graph is a breakdown of what's in that video. Each of these lines corresponds to a section of the video and the length of each line corresponds to the length of time that particular section takes. So these lines indicated here in yellow with dots on their edges represent content original to the New York Times video, either clients speaking directly to camera or some visualization that appears to have been created inside the New York Times. The rest of the lines in this graph represent segments of the video, making up roughly a third of the runtime that come from other sources. What other sources did they pull this from? Damn to find out, that's the problem.
I've already given you more links to sources on the screen in the video you are watching now than Ezra Klein and his entire team bothered to give the public and their entire almost 30 minute piece. Roughly nine minutes of included video clips and screenshots are in that video, and zero links are given. No links to sources in the description, none in the transcript. I even searched the New York Times site until I found their page for this video, and I looked at the transcript on their page for it. No links or URLs given there either. Not just that, but the reason that I had to search the New York Times site for its transcript is because, as you can watch here, at the time that I'm recording this video, the link to the Times web page for the transcript, as given in the video's YouTube description, goes to a broken 404 page. The single solitary URL in the YouTube description that does work is a marketing link that goes to the book recommendation page for Ezra's podcast where he interviews people.
Instead of links, here is what passes for attribution at the New York F-ing times. These three screens. Here's one. Here's the second. Here's the third. Now, I'm going to stack them all up so you can see them all at once. Somewhere in this 29 and a half minute video, according to the New York Times, there's something that they got from Anthropic, something that they got from Google, and something that they got from OpenAI. Where in the video were those things? Where exactly on those sites did they come from? F-You, that's where. Three AI companies, four TikTok channels, eight news organizations, four YouTube channel, a bunch of stock video photo providers, one sub stack, one Patreon, six Twitter slash X users, and a few other organizations. And they don't even seem to be in any particular order. The video starts with what appears to be an Anthropic ad followed by a gym and I add. And then instead of going to OpenAI, as you might think from looking at this, it shows three, not four, but only three TikTok clips, then back to Ezra, then to a CNN clip followed by BBC one CBS doesn't even appear until later. So if someone who, unlike Ezra, actually cares about the integrity of where their information comes from, wants to break down for people, the numerous fallacies and conflicts of interest in what Ezra is saying here, the amount of work that has to be done is enormous.
### 意见专栏不是免责借口：深度伪造时代的核查基准
有人可能会辩解称这只是意见专栏（Opinion column）而非硬新闻报道，因此不必苛求严格的引用规范。然而这种辩解完全站不住脚：当手握巨大话语权与公共平台的媒体名人公开倡导某种政策立场时，他们绝对有义务公开其观点所依赖的证据链。况且，这种不透明绝非专栏特例，《纽约时报》在关于美国国家气象局使用 AI 的常规新闻报道中同样如此——没有 URL、没有超链接，只有一份模糊的来源机构名单。
在技术层面，为引用的内容添加链接极其轻而易举。一个仅由单人运营的独立频道尚且能坚持数年为所有论据附上出处，而配备了三名全职事实核查员（Fact checkers）的专业制作团队却毫无作为。随着 AI 技术的演进，这种透明度的缺失将带来系统性灾难：
1. **深度伪造门槛骤降**：无须等待假设中的通用人工智能或超级智能，现有的**生成式人工智能**（Generative AI: 能够生成逼真文本、图像及视听内容的机器学习系统）已经完全有能力制造以假乱真的音视频。
2. **针对性认知作战**：无国界记者组织（Reporters Without Borders）的报告记录了上百起利用 AI 深度伪造冒用记者身份的案例；仅在得克萨斯州的地方选举中，便已经出现了多起针对候选人的 AI 虚假政治广告。
3. **权威背书的链条断裂**：公众必须拥有核验权威言论真实性的渠道。如果任何政客、CEO 或专家的表态视频都无法追溯原始出处，信息空间将彻底失去验证机制。
Original English Source
Quick aside, some of you might think that I'm being unreasonable because this is an opinion column and not a news story. First off, no, I'm not. When media personalities with huge platforms advocate for policy positions, which is exactly what Ezra is doing here, they absolutely owe us references to what they're basing their opinions on. And even if I was being harsh, which I'm not, here's a link to an actual times news story on the recent increase in use of AI by the national weather service. And it's no different. No links, no URLs, just a list of companies and channels that might have been the source for some piece of footage or some image somewhere in the video.
Make no mistake, adding links to the sources you use is easy to do. Hell, I do it. I've done it for two and a half years and I'm just one person. Ezra Klein's video has three fact checkers credited on it. It would be absolutely trivial of them to include the links if they cared at all. And it's never, ever been more important. And it will be getting more and more important over time. Why? Because AI makes creating disinformation so easy and so hard to spot. I'm not talking about some pipe dream sensationalist, Doomer, fantasy, super AI either. The current state of AI can absolutely do this. In fact, it's been doing it for years and it doesn't require any tech advances at all.
So here's a link. See, it's not that hard. Here's a link to a report from Reporters without Borders. About 100 instances where journalists were impersonated with AI deepfakes. And that report only goes up to 2025. And here are two more links to stories about two separate deepfake political ads targeting two different candidates for next month midterm elections, just in Texas where I live, just in state or addresses. For months now, it has been trivial to make deepfakes of news personality, saying whatever you want them to see this time's report on deepfaking Sam Altman. And even in the event that AI never gets any closer to hypothetical super intelligence than what we have now, making deepfakes will almost certainly continue to get much easier.
This is why it's critical that we demand that the media accurately reports where they got their references and sources. At a bare minimum, we need to be able to trace the origin and authenticity of every clip or quote that purports to be a statement from anyone in a position of power or authority, be they CEO, politician, journalist, subject matter expert or anyone else whose position or expertise is relevant to the public discourse.
### 布兰多里尼定律与新闻业职业标准的衰退
在信息的传播博弈中，网络上著名的**布兰多里尼定律**（Brandolini's Law: 驳斥虚假信息所需的能量要比制造它大一个数量级，又称废话不对称原理）深刻揭示了这一困境。主流媒体松懈的标准，正是导致辟谣工作举步维艰的核心元凶：
* **资源分配的结构性失衡**：真正关注真相的个体与独立创作者必须付出高昂的时间和精力去追索源头并记录证据；而手握庞大资源与预算的机构媒体却肆无忌惮地用点击诱饵淹没受众，甚至不屑于展示其论证过程。
* **专业标准的全面倒退**：享有丰厚报酬的《纽约时报》专业采编人员，在信息透明度与引文规范上的自我约束，甚至远低于维基百科（Wikipedia）上不计报酬的志愿编辑。
当一家标榜行业最高水准的新闻机构连最基础的透明度和引用规范都弃之不顾时，这不仅是 Ezra Klein 或某一个专题团队的失职，更是整个美国新闻业的羞耻。在深度伪造与合成媒体泛滥的新现实下，公众必须通过舆论与反馈机制对传统媒体施压，重构一套不再令行业蒙羞的信源公开标准。
Original English Source
There's this thing on the internet called Brandolini's law, which dates more or less. The amount of energy needed to refute disinformation is an order of magnitude bigger than that needed to produce it, except they use a more profane word than disinformation. The lax standards of the media is one of the big reasons why debunking B.S. is so much harder than generating it. Because those of us that actually care about the truth and hold ourselves to a basic standard of transparency have to go to so much more effort to track down and document the actual sources we use than they do, and the large companies like the Times spend their vastly superior resources overwhelming the public with clickbait and not bothering at all to show their work.
Ezra Klein and the New York Times should be ashamed of themselves. And if you agree, I'd urge you to comment on their YouTube videos and let them know. And furthermore, especially if you're in New York Times subscriber, here's one final link for you. It's a form where you can tell the New York Times that they have no business calling themselves "the most powerful engine for independent boots-on-the ground and deeply reported journalism" when their paid staff isn't even held at the same standards as unpaid volunteers on frickin' Wikipedia. They want to claim to be setting the standard. Well, it's past time that we had a standard that was no longer embarrassing to the whole of American journalism. Thanks for watching. Let's be careful out there.