Internal AI Safety Concerns and Executive Divergence

Anthropic's alignment science lead estimates a 10% chance of human extinction from AI within a decade. This claim follows the resignation of a researcher who argues current industry practices lack responsibility regarding self-improving superintelligence.