Unaligned AI

Rapid progress in artificial intelligence is outpacing reliable alignment and control methods, driven by competition among labs and nations. Surveys of AI researchers show a median estimate of about 10 percent that advanced AI causes human extinction or permanent disempowerment. [[1]](https://aiimpacts.org/wp-content/uploads/2026/09/ESPAI2024.pdf) Researchers at leading labs have stated probabilities above 10 percent of AI killing all humans within a decade. [[2]](https://www.theguardian.com/technology/2026/sep/09/ai-superintelligence-risks-warnings-scientists-politicians) These exceed estimates for nuclear war or extreme climate change, placed near 0.1 percent each over the century. [[3]](https://www.ft.com/content/cb263cd8-b63d-4382-958b-fd06ee4f120e?syn-25a6b1a6=1) Environmental and geopolitical risks can cause vast harm, yet a misaligned advanced AI uniquely threatens irreversible loss of control or extinction through speed, scale, and independent goal pursuit.