Safety researcher: current models already pose grid-down cyber risks to infrastructure
gmiller · x · 2026-11-16
AI safety researcher gmiller warns that even short of full existential risk, today's flawed and limited models can already impose major cyber risks on human infrastructure. That's one big reason he's worried about imminent recursive self-improvement — not because it causes extinction within 18 months, but because grid-down level risks could arrive soon.
Related event: AI Safety Researcher Warns Current Models Pose Grid-Level Cyber Threats(2 posts)→
More from AGI Musings
- Asimov Press founder: academic journal innovation is strangled by indexers like Clarivate — NikoMcCarty · 2026-11-16
- Gary Marcus: I respect honest doomers, not those who make things up about me — GaryMarcus · 2026-11-16
- Only 5% of 28,000 US taxpayer-funded patents licensed in 1980 — an analogy for AI training data rights — krishnan · 2026-11-16
- Chollet: early-2026 "SaaSpocalypse" panic was BS, Stripe data shows — fchollet · 2026-11-16
- Gary Marcus tells critics to watch his TED talk: 1.6M plays on AI risks — GaryMarcus · 2026-11-16
- Sebkrier: Bay Area AI Pessimism Is Crushing, but Foresight Institute Feels Like 90s Tech Optimism — sebkrier · 2026-11-16