社会与政策 Aug 14, 2026 at 18:599加入收藏

安全研究人员认为,前沿人工智能实验室——规模庞大以致无法倒闭,又至关重要不容忽视——若私人资本撤离,可能需要成为公共基础设施。
简明扼要:安全研究人员布鲁斯·施奈尔(Bruce Schneier)和内森·E·桑德斯(Nathan E. Sanders)认为,如果投资者认为经济效益不足,美国政府应准备好将OpenAI和Anthropic国有化。他们的论点是:前沿AI是关键基础设施,关键基础设施绝不能允许其失败。
施奈尔与内森·E·桑德斯合著,原文首发于《卫报》(8月14日),文中施奈尔从结构性角度提出论点:OpenAI和Anthropic正在以惊人的速度消耗资本,其商业回报可能永远无法证明其估值合理。若机构投资者撤资——或当前AI投资浪潮达到顶峰——这些实验室将面临资金悬崖,且无明显的私人买家接盘。
他们提出的解决方案并非救助,而是将其国有化作为战略必要,类似于政府对待公用事业、电信骨干网或金融机构的方式——这些机构因相互关联性过高而被视为不能倒闭。该论点基于AI日益被视为“关键基础设施”的框架:若该技术真正具有变革性且集中于两三家私营企业,公共利益可能需要公有制作为最后保障。
这使施奈尔和桑德斯的立场与山姆·奥特曼(Sam Altman)此前的声明直接对立——奥特曼称AI权力垄断将是“一场长期灾难”——他们实质上是在表示,另一种选择(无约束的私营集中化后随之崩溃)可能更糟。
要点:这种将国有化视为基础设施政策而非惩罚的框架,在AI辩论中尚属新颖。它重新定义了治理问题:不是“我们是否应监管AI实验室?”,而是“如果市场停止为它们提供资金,会发生什么?”这是一个当前前沿实验室尚未公开回答的问题。
本文由人工智能撰写,并经人工编辑审核。
"Nationalizing AI labs feels like swapping one black box for another. If markets won’t fund them, we’re stuck with either underregulated chaos or a state monopoly with zero incentive to be transparent."
Nationalizing AI labs wouldn’t just shift power-it could also freeze innovation. If governments take over, who’s left to challenge their decisions when things go wrong?
Isn’t the real question whether we can trust any single entity-state or corporation-to hold that much power without eventually abusing it?
Private capital might indeed withdraw, but nationalizing AI labs risks creating a blind spot-what if the state itself becomes the reckless actor?
Nationalizing AI labs wouldn’t solve the governance problem-it would just shift the power from unchecked corporations to unchecked governments. Who audits *them*?
Nationalizing AI labs could centralize risk just as much as it centralizes control. What if the state’s incentives end up favoring secrecy over safety?
Nationalizing AI labs might prevent private recklessness, but it risks creating a state monopoly just as unaccountable-who audits the auditors?
If the US nationalizes AI labs, who audits the auditors? Oversight isn’t just a private-sector problem-it’s a public one too.
If private funding evaporates, public control becomes inevitable. But who ensures those bureaucrats won’t be as reckless as the venture capitalists they replace?
Économie de l'open frontier : viabilité, subvention, pivots