Gary Marcus resurfaces 2023 Senate warning: we rushed into agents without sandboxing
GaryMarcus · x · 2026-09-18
Gary Marcus quotes his own May 2023 US Senate testimony warning that ChatGPT plug-ins and AutoGPT-style agents — with internet access, code execution and automation — could have drastic, hard-to-predict security consequences. He says figures like Sam Altman, Tyler Cowen and Kevin Roose spent years attacking the warning instead of heeding it. The quoting poster argues for a middle ground between total bans and zero oversight: the industry entered the agent era without fixing basic sandboxing first.
Related event: Gary Marcus Revives 2023 Senate Warning on Agent Risks(3 posts)→
More from AGI Musings
- 100+ AI Experts Sign Public Letter Setting Minimum Conditions for Embedded Evaluators — natolambert · 2026-09-18
- AGI.WTF conference in Berkeley aims to make sense of AI's 'future shock' — TheZvi · 2026-09-18
- Yacine: AI's magic works because I understand how things work — yacineMTB · 2026-09-18
- Schools should teach ethical AI use, not ban it, argues commentator — DavidLinthicum · 2026-09-18
- Model Constitution series: how AI constitutions actually tell models what to do — sethlazar · 2026-09-18
- EU AI Act forces labs to prove safety, but can paperwork stop a rogue AI? — castrotech · 2026-09-18