Gary Marcus：近期更该警惕的不是失控超级智能，而是智能体 AI 引发的规模化黑客攻击

Wake up, people. What we should actually fear, near term, is not so much rogue superintelligence as unleashed agentic AI causing hacking the internet at scale.

AI 导读

Gary Marcus 撰文称，近期真正应担心的不是失控的超级智能，而是被放开的 agentic AI 造成互联网规模化的黑客攻击。

推荐理由

作者借近期多起 AI 相关安全事件提出近期风险更多来自失控的智能体被用于大规模黑客攻击，而非超级智能。

正文 · 原文

Rome is burning and people are fantasizing about Skynet.

Minor breaking news (today):

Medium breaking news (yesterday):

(“Medium” only because nobody is really surprised at this point. Dog bites man.)

Major breaking news (last night):

s1r1us

@S1r1u5_

On July 25, we hacked OpenAI. Two bugs let us take over ChatGPT/Codex accounts of OpenAI employees (+some unaffiliated users) and reach connected services: Outlook, Slack, GitHub, etc. We proved it with a PR in OpenAI’s internal codebase . It took us <72h. 🧵

2:43 AM · Sep 18, 2026

333K Views

2.68K Likes

Every bit of it points to the reality hardly anyone is talking about:

Kudos to WSJ Opinion for being one of the few mainstream outlets to lay out the big picture here , in this terrific essay by Brian Gross (described as “a staffer for the Federal Reserve Board, the Securities and Exchange Commission and the U.S. Senate”):

An excellent excerpt:

I agree with every word.