Author calls for freezing frontier AI to handle harms already here, in new book Obsolete
GarrisonLovely · x · 2026-10-20
Garrison Lovely's new book Obsolete (OR Books) argues for freezing frontier AI development to address harms already occurring. He predicts AI-enabled cyberattacks will worsen over the next two years, citing Anthropic's Mythos—reportedly the first AI to discover and exploit vulnerabilities better than human experts. The book challenges the bubble-vs-utopia false binary.
More from AGI Musings
- Stanford's TRACER: Training-Free Multimodal AI Tracks Hospital Patient Behavior — StanfordAILab · 2026-10-20
- Dev asks: AI coding killed the satisfaction of finishing a project — Individual_Lime_1626 · 2026-10-20
- Is 'Eval Awareness' in AI Models Emergent — or Trained In by Feeding It Studies? — bendee983 · 2026-10-20
- Nature essay argues current AI is extremely unlikely to be conscious — intelligence alone isn't enough — DeryaTR_ · 2026-10-20
- Incremental model updates are overrated — the real leap is 10M-token context windows — gordriver_berserker · 2026-10-20
- Victoria Krakovna ships major update to her AI alignment resources list — vkrakovna · 2026-10-20