State AG lawsuits target AI loss-of-control risk, Arkansas seeks to bar recursive self-improvement
ghadfield · x · 2026-10-18
AI safety researchers Justin Bullock and Nathan Calivn flag that state lawsuits from Arizona AG Mayes and other state AGs deserve more attention — reporting has focused on kids' safety, but the complaints go further.
The Arizona complaint argues that "the threatened harm—loss of human control over autonomous systems capable of attacking private, governmental, and critical-infrastructure networks—is imminent, substantial, is not adequately remediable by money damages." The Arkansas lawsuit explicitly asks the court to enjoin defendants from creating models capable of recursive self-improvement, showing AGs are formally writing loss-of-control and RSI risks into legal demands.
More from Safety
- One diagram explains the difference between CSRF and XSS attacks — _jaydeepkarale · 2026-10-18
- Grady Booch slams Anthropic for tolerating up to 50 deaths and $1B damages per model — Grady_Booch · 2026-10-18
- Single Neuron Suffices to Bypass Safety Alignment in LLMs, Paper Finds — burny_tech · 2026-10-18
- Polymarket gives OpenAI just 8% odds of a full training pause by end of 2026 — Polymarket · 2026-10-18
- OpenAI engineer over alleged hidden recording device seeks bail, denies Saudi escape plan — Polymarket · 2026-10-18
- Free comic "The Collective" offers a strange retelling of the Hugging Face hack — Flugelwagen · 2026-10-18