Could humans become AI's "physical sub-agents"? Cybersecurity seen as humanity's key defense
Michael_D_Moor · x · 2026-09-27
MichaelDMoor argues that one of the biggest near-term AI safety risks is bad actors using next-gen models to hack and extort individuals at scale, turning them into cheap and effective "physical sub-agents" — potentially more practical than robots. He sees cybersecurity as possibly the single most important frontier for protecting humanity, with going fully analog as a personal defense.
More from AGI Musings
- Oxford's Bronstein: time to put the "stochastic parrots" argument to rest — mmbronstein · 2026-09-28
- MIT's David Bau: X comments drowned in AI slop clichés — or are humans just talking like AI now? — davidbau · 2026-09-28
- Astro Teller on moonshots: attack the hardest assumption first, kill zombie projects — PeterDiamandis · 2026-09-28
- Prediction: coding will surpass all humans within 4-6 months, making engineering experience worth zero — morganb · 2026-09-28
- "Give them usage resets and AI slop and they will never revolt" — arpitingle · 2026-09-28
- Einstein didn't reweight hypotheses—he changed the hypothesis space itself — vishalmisra · 2026-09-28