AI Safety Researchers Debate Whether Frontier Model Training Should Continue

Safety researchers Rohit Krishnan and David Manheim clashed over whether frontier model training should continue, with Krishnan arguing current methods need not be perfect today, while Manheim countered that unprecedented disasters are no reason to dismiss risk.

2026-10-14 ~ 2026-10-14 · 3 related posts

• Rohit Krishnan on frontier risk: just make Fable 7 better, and there's no Fable 8 if it isn't — krishnanrohit · 2026-10-14
• Safety researchers clash: does failure rate need to approach zero as stakes go infinite? — davidmanheim · 2026-10-14
• David Manheim: Every AI Catastrophe Will Be 'Unprecedented' — Until It Repeats — davidmanheim · 2026-10-14