Gary Marcus: I Warned Four Years Ago About LLMs Hooked to Power Grids — Now We're Here
GaryMarcus · x · 2026-09-18
Gary Marcus doubles down on his early warnings: in Senate testimony three years ago he said that a month after GPT-4's release, OpenAI shipped ChatGPT plug-ins, quickly leading others to build AutoGPT — with direct internet access, code writing, and increased automation powers, likely causing drastic, hard-to-predict security consequences. He argues that's exactly where we are now. He had also warned about hooking unreliable LLMs to consequential systems like power grids, driverless car fleets, and air traffic control, and pushes back on the idea that only "doomers" issued warnings.
More from AGI Musings
- Anil Seth reflects on two years of debate over his conscious-AI target article — anilkseth · 2026-09-18
- Anil Seth's Conscious AI and Biological Naturalism collection out in BBS with fifty commentaries — anilkseth · 2026-09-18
- Anil Seth's conscious AI collection published in BBS with 50 commentaries — anilkseth · 2026-09-18
- Noam Brown: models may perform their chain of thought; alignment must be solved — infoxiao · 2026-09-18
- The 'Seniority Cliff': skipping junior-level friction may hollow out engineering intuition — Jumpy-Increase9337 · 2026-09-18
- Gary Marcus: the 'nobody saw AI security risks coming' narrative is completely false — GaryMarcus · 2026-09-18