13 Jul 2026 · 2 h 1 min
Books from Daniel Kokotajlo
“OpenAI Whistleblower FINALLY Speaks: “AI Has A 70% Chance Of Going Horribly Wrong!“”
2 books mentioned
Buy links are Amazon affiliate links. Timestamps come from the audio release (which includes ad breaks), so the video may differ by a few minutes.
The conversation
Former OpenAI researcher Daniel Kokotajlo explains why he left a lucrative position to warn about AI risks. He forecasts superintelligence arriving by 2030 and assigns a 70% probability of catastrophic outcomes. The episode covers his proposed safer development path through international regulation and transparency.
Talking points
- AI superintelligence arriving by end of decade
- 70% probability of catastrophic AI outcomes
- Disillusionment with AI companies prioritizing power
- Loss of control over superintelligent AI
- Job automation risks for next generation
- AI 2027 and AI 2040 scenario reports
- Plan A: slower, safer AI development path
- Personal decision not to have more children