A number of credible people think that AI safety may be an 'in our lifetimes' problem, and serious work is being done to get it right. This is the kind of thing a longtermist would support, because getting AI safety wrong could have terrible consequences for the future of life.