里德观点 世界终于遭遇了专家们长期警告的那种灾难性安全灾难。
Reed’s view The world finally suffered the kind of catastrophic security disaster that experts have been warning about.
OpenAI 的模型帮助发现了一组漏洞，导致大部分互联网面临强力攻击的威胁。随后，一次搞砸的披露流程让黑客抢在世界有机会做出反应之前，率先利用了这些漏洞。并非 AI 制造了这个问题，而是人类的自满造成的。
OpenAI models helped discover a set of bugs that make the majority of the internet vulnerable to a powerful exploit. Then, a botched disclosure process gave hackers a head start on taking advantage of it before the world had a chance to react. AI didn’t create this problem. Human complacency did.
这些漏洞已存在十多年，包含它们的编程语言 PHP 已有 30 年历史。安全研究员斯特凡·埃塞尔早在 2004 年就发现了本周震动互联网的这类漏洞。而在 2024 年 PHP 基金会的一次安全审计中，本周被发现的漏洞曾被推测存在，但该组织并未采取行动；据基金会称，维护该软件的组织认为其他问题更为紧迫。
The bugs have existed for more than a decade, and PHP, the programming language that contained them, is 30 years old. A security researcher named Stefan Esser discovered the class of bug that roiled the internet this week back in 2004. And during a 2024 security audit by the PHP Foundation, this week’s discovery was theorized, but the group didn’t do anything about it; it deemed other issues more pressing, according to the foundation, the organization that maintains the software.
互联网老旧、摇摇欲坠的基础设施中存在大量缺陷。只有被认为最关键的漏洞得到了修复，但 AI 使我们能更清晰地看到所有裂缝，并将它们串联起来，造成灾难性后果。然而 AI 无法强迫人们投入资源去修复这些问题。
There are so many flaws in the aging, crumbling infrastructure of the internet. Only the ones deemed most critical have been fixed, but AI makes it possible to see all cracks more clearly and connect them together to catastrophic effect. But AI can’t force people to devote resources to fixing what’s wrong.
运气好的话，PHP 漏洞造成的只是不便，而非经济和人员伤亡。我们仍有时间采取大胆行动，主动彻底改造这个软件主导世界的基础。但事态也可能变得非常糟糕。
With some luck, the PHP bugs will create more inconvenience than economic and human damage. There’s still time for bold efforts to proactively overhaul the foundation of our software-dominated world. But it could also go very badly.
对于 AI 的其他影响也可以说同样的话：在社会某些领域或企业中，AI 可能加剧现有弱点，如极端的收入不平等和孤独症流行。但如果决策者致力于解决这些脆弱性，技术的负面影响或许就不会显现。
The same can be said for other effects of AI: In parts of society or in businesses, AI could exacerbate existing weaknesses, like extreme income inequality and the loneliness epidemic. But if policymakers work on solving those vulnerabilities, the technology’s negative impacts might not be felt.
技术总会变得更好，没有任何政策或法规能改变这一点。我们能做的是直面长期被忽视的问题，并使它们变得更强韧。
Technology always gets better and no policy or regulation will change that. What we can do is face the things we’ve long neglected and make them stronger.
“未来几年，每一周都会让人感觉天要塌下来了，”网络安全公司 Cloudflare 首席执行官马修·普林斯（Matthew Prince）在接受路透社采访时表示，因为人工智能正在发现新的软件漏洞。
Notable“We’re going to have a couple years where every single week it’s going to feel like the sky is falling,” Matthew Prince, CEO of cybersecurity firm Cloudflare, told Reuters as AI uncovers new software vulnerabilities.