ARTICLE · 1024219
AI巨头的“特洛伊木马”:科技巨头渲染AI末日呼吁监管实为利己
AI巨头的“特洛伊木马”:科技巨头渲染AI末日呼吁监管实为利己Under cover of warning against the AI apocalypse, top AI companies are blatantly seeking special privileges: That’s the best explanation for the sudden flurry of doom‑mongering from CEOs like Anthropic’s Dario Amodei and OpenAI’s Sam Altman. These moguls say their industry needs to tap the brakes on the tech’s rapid progress, but will warn that the United States still needs to beat China in the AI‑development race — leaving a de‑facto cartelization as “the only choice,” with the feds setting up guardrails and offering their companies protections against disastrous mistakes. Firstly, he showed up knowing he was not welcome. Second, he had a little laugh fest with his comrade, Rep. Alexandria Ocasio‑Cortez. That is: What they really want is to protect themselves from liability by shifting responsibility to the government. Putting the AI giants in great positions for their coming IPOs, this would protect Anthropic, etc., if their AIs go rogue, causing billions or trillions of dollars in damage, just as “too big to fail” protected the megabanks that bumbled their way into the 2007‑08 global financial crisis. Amodei also advocated in a blog post for Washington to give frontier labs a so‑called antitrust waiver, which would allow them to coordinate with one another without raising federal collusion or monopoly concerns — essentially a pass for companies to regulate themselves — to ensure safety. Now, you don’t need an antitrust waiver to make sure technology is safe, as former FTC Commissioner Alvaro Bedoya pointed out. You do need one if you’re looking to coordinate on other areas beyond keeping people safe. Any waiver, should be met with deep skepticism in light of the economics of the industry and the threat they face from open‑source models. In some respects, I can’t blame them for taking this approach. These companies have burned hundreds of billions in development and are still losing money — OpenAI lost approximately $39 billion in 2025 — so they depend on hitting record‑setting values in their IPOs. And while their technology may be used to cure cancer or infinitely increase productivity, the companies themselves won’t necessarily profit from such innovations. Maybe they are sincere in their belief AI could doom the world, but it’s awfully convenient to push for regulation when they are on the cusp of cementing their role as worldwide leaders. Anthropic’s public prospectus is expected within weeks and could value the giant at $2 trillion. OpenAI pulled their IPO plans, with reports suggesting they will now go public in 2027. The whistleblower behind the latest panic is also suspect. Jacob Coxon — the 27‑year‑old researcher who quit Anthropic after just four months, following three years at OpenAI — went viral warning the companies were “gambling with our lives.” His post racked up more than 100 million views on X, despite the fact he had almost no prior history on the platform. And within the next few days he had landed wall‑to‑wall coverage and interviews with Fox, CNN, NBC, ABC, The New York Times and The Wall Street Journal. Nothing odd about a young nerdy researcher with no known media experience dominating an entire news cycle. On Tuesday, he is slated to appear at the Future of Life Institute’s “Pro‑Human Assembly” alongside figures such as Steve Bannon, Bernie Sanders and Randi Weingarten. Really? That’s Coxon’s solution after warning that AI could kill us all? Bring in a man old enough to be his grandfather and the widely reviled head of a teachers union to devise a plan? I could hardly conceive of a more bizarre, not to mention feckless, group to create a framework for reining in the most advanced technology known to mankind. The best take was perhaps Clement Delangue, the CEO of AI company Hugging Face, who claimed asking Coxon “about AI extinction risk is like asking your AC guy about climate change.” Amid all this AI doom and gloom, you have to ask — who stands to benefit? The answer is, of course, incumbent frontier labs and our foreign adversaries who will happily race ahead if we slow down. Fortunately, the White House isn’t playing into it. JD Vance said frontier AI companies “begging the government to regulate them” felt like “a Trojan horse.” Trump, meanwhile, explained, “We’re leading China in AI. We’re the most sophisticated country in the world, and frankly I want to keep it that way because whoever wins AI wins.” It is fair to have questions and concerns about AI, its bias and its impact on people. But maybe the best place to get answers isn’t always from the people who are running it, especially when rivals like Amodei and Altman seem suspiciously eager to fall on their swords. As Trump put it: “When, in the History of Business, did anyone see the Leaders of an Industry call for Regulation that, if strongly implemented, will drive them into oblivion and bankruptcy? AI, and Data Centers, will be the Greatest Economic Development Engine in History.” 中文翻译 在渲染AI末日危机的幌子之下，头部AI企业明目张胆谋求特殊特权。这也正是Anthropic首席执行官达里奥·阿莫代伊、OpenAI山姆·奥尔特曼等高管集体散布末日论调的合理解释。 这些巨头声称行业必须放缓技术狂奔的脚步，同时又警告美国仍需要在AI竞赛中击败中国。于是在他们口中，由联邦政府设立监管护栏、为本企业规避灾难性事故风险，形成事实上的行业卡特尔，就成了“唯一出路”。 他们真实诉求是转移风险，让政府替企业承担责任，从而为即将到来的IPO铺平道路。一旦AI系统失控造成千亿、万亿美元级别的损失，这套保护机制就会生效，如同“大而不能倒”原则庇护2007‑2008金融危机中闯祸的大型银行。 阿莫代伊还在博客中呼吁华盛顿给前沿AI实验室发放所谓反垄断豁免，允许企业之间协同行动，不会被认定串通或垄断，实质上等于允许企业自我监管。前联邦贸易委员会委员阿尔瓦罗·贝多亚指出：保障技术安全根本不需要反垄断豁免；只有想在安全之外开展协同，才需要这份豁免。结合行业经济现状、开源模型带来的竞争压力，任何此类豁免都应当遭到高度质疑。 某种角度上，我并非完全不能理解企业的动机。这些公司烧掉数千亿研发资金，至今仍持续亏损。OpenAI仅2025年亏损约3900亿美元，企业只能寄希望于IPO获得创纪录估值。即便技术未来可以治愈癌症、大幅提升生产力，企业未必能从中盈利。也许他们真心相信AI会毁灭世界，但恰恰在即将坐稳全球行业龙头之际呼吁监管，时机太过巧合。Anthropic几周内就要公开招股书，估值预计可达2万亿美元；OpenAI搁置IPO计划，报道称计划2027年上市。 本次引爆恐慌的爆料人本身疑点重重。27岁研究员雅各布·考克森，在OpenAI任职三年，仅在Anthropic干了四个月就离职。他发布帖子警告企业“拿所有人的性命赌博”，在X平台播放量破亿，而他此前几乎没有任何平台发声记录。短短数天，福克斯、CNN、NBC、ABC、《纽约时报》《华尔街日报》全部对他进行专访。一个没有媒体从业经验的年轻技术人员，霸占完整新闻周期，这件事本身就很反常。周二他将出席生命未来研究所的“亲人类大会”，同台嘉宾包括史蒂夫·班农、伯尼·桑德斯、兰迪·温加滕。这就是他口中防止AI毁灭人类的解决方案？找来年龄足以当他祖父的政客，以及饱受争议的教师工会领袖制定约束人类顶尖科技的规则，很难想象还有比这更怪异、更无力的组合。 AI企业Hugging Face首席执行官克莱门特·德朗格的点评一针见血：“向考克森询问AI灭绝风险，就如同找空调维修工咨询气候变化。”铺天盖地的AI末日论调之下，要追问：谁会从中获益？答案显而易见：占据市场的头部前沿实验室，以及一旦美国放慢脚步就会全力追赶的境外对手。所幸白宫并未落入圈套。万斯表示，前沿AI企业“乞求政府监管自己”，本质就是“特洛伊木马”。特朗普表示：“我们在AI领域领先中国，我们是全球技术最顶尖的国家，我希望维持这个优势，赢得AI竞赛者将赢得未来。” 我们完全有权质疑AI算法偏见、社会影响。但答案未必总该由运营AI的企业给出，尤其阿莫代伊、奥尔特曼这些竞争对手，表现出一种可疑的自我牺牲姿态。正如特朗普所言：“商业史上，你何曾见过一个行业的领袖主动呼吁监管，而严格执行监管本可以让他们走向覆灭破产？AI和数据中心，将会成为人类历史上最强劲的经济发展引擎。”