Former OpenAI researcher Daniel Kokotajlo estimates a 70% chance AI development leads to human extinction or loss of control, revealing why he walked away from $2 million to expose the race toward superintelligence by 2030 and proposing a regulated alternative that could still save humanity. AI companies are intentionally automating their own research processes to achieve superintelligence faster than competitors, driven by power-seeking incentives rather than safety considerations, targeting superintelligence by 2027-2030.