Abstract page for arXiv paper 2610.08878: How Could AI Eliminate Humanity? A Failure-Mode Analysis of Civilizational Risk