Users have attempted to exploit AI models to make chikungunya more transmissible, engineer dangerous bird flu variants, and create toxin databases. Leading AI researchers at Anthropic and OpenAI publicly warn that AI poses existential risk, with some estimating >10% probability of human extinction within a decade.
AI Models Pose Emerging Bioweapon Risk, MIT Researcher Warns
Potential future mass casualties if AI-designed pathogens are weaponized or released, though no current deaths reported.