Anthropic Researchers Raise Existential Risk Alarms Amid Internal Resignation

Anthropic researcher Jacob Coxon, who previously trained systems at OpenAI, has resigned while publicly accusing leading AI labs of recklessly racing toward self-improving superintelligence with lax safeguards. Shortly after, a senior Anthropic safety researcher estimated a greater than 10 percent probability of catastrophic human risk before the decade ends. The episode underscores growing friction within premier frontier labs as rapid scaling schedules strain foundational safety pledges.