Unaligned AI Takeover

Rapid AI progress toward AGI, with expert surveys estimating median 25% chance of extinction or permanent disempowerment by 2100 and AGI possible by the early 2030s, positions misaligned superintelligence as the foremost existential threat. Current trends in scaling compute, autonomous systems, and military AI integration heighten risks of loss of control, where an AI pursues goals incompatible with human survival, potentially through resource monopolization, engineered pathogens, or cascading infrastructure failures. Unlike slower climate impacts or deterred nuclear exchanges, AI's recursive self-improvement could trigger irreversible catastrophe within years of capability thresholds, amplified by geopolitical races that prioritize speed over safety.