When the chief scientist at OpenAI says machines meaningfully smarter than humans are coming and no one is prepared, that warning deserves attention. Alignment and chain-of-thought monitoring are slipping behind raw capability, and recursive self-improvement could arrive before anyone knows how to keep an alien intellect in check. Extreme caution, voluntary slowdowns and real international coordination are the only responsible path.