Gary Marcus: infinite AI bot hack armies, not rogue AI, are the near-term disaster

GaryMarcus · x · 2026-09-18

Gary Marcus warns the near-term AI disaster isn't rogue superintelligence but something already feasible: with enough money, you can unleash an unlimited number of bots to endlessly probe every published vulnerability on any server until one breaks in. His points: it's basically already happening, needs no RSI or speculative scenarios, we're doing nothing to stop it, and it will be bad. A quote-tweet notes such activity would have meant jail a decade ago.

Related event: Gary Marcus: The real near-term AI threat is bot hacker armies, not superintelligence(4 posts)→

Original post →

More from AGI Musings

AGI Musings channel →