Stanford Researcher Calls for Heavy Penalties on AI Labs Over Recent Model Releases
anshulkundaje · x · 2026-09-15
Stanford researcher Anshul Kundaje and antibody42 debate AI release accountability:
- antibody42 argues several companies have effectively published evidence of felonies in recent months, and that punitive laws only bite after a sufficiently damaging lesson.
- Kundaje agrees recent (relatively benign) incidents deserve heavy penalties, which would push labs toward real due diligence before release — effectively a slowdown for technical reasons.
- He rejects slowdown theater: if internal and external evaluators are confident on alignment and deception risks, models should ship with full liability, which itself acts as a pressure point.
More from AGI Musings
- What do we call the 'it was ever thus' rhetoric used to dismiss AI concerns? — jjvincent · 2026-09-16
- AI's hardest problems need democratic deliberation — and independent experts — RishiBommasani · 2026-09-16
- Continual Learning in Fruit Fly Brain Decoded, Framed as Missing Piece for AGI — skolnaja · 2026-09-16
- Pedro Domingos: AI is anti-moat, dissolving switching costs that protect IT providers — pmddomingos · 2026-09-16
- Founder pushes back on Anthropic CEO's runaway-AI warnings on NDTV Profit — angadc · 2026-09-16
- AI Safety comms debate: punchy messaging wins short-term but erodes community epistemics — NathanpmYoung · 2026-09-16