Polymarket: Only 10% odds US enacts AI safety bill by end of 2026

Polymarket · x · 2026-11-23

Polymarket's prediction market on US AI safety legislation shows just 10% odds of a bill becoming law by end of 2026, but odds rise to 63% by June 2027 — the market sees legislation as likely, just not soon.

The market defines a qualifying bill strictly: it must include at least one of — a ban on releasing specific models or mandated federal review, training restrictions (data/compute/parameter caps), usage prohibitions, or human-in-the-loop requirements. Voluntary review submissions don't count.

Context adds that OpenAI's Chief Scientist warned AI could trigger an "avalanche" of research breakthroughs as early as 2027, outpacing current safety measures.

Original post →

More from AGI Musings

AGI Musings channel →