ohsecurity··on AI overly affirms users asking for personal adviceNot that surprising. If you optimize for a pleasant interaction, you often get agreement instead of correction. The question is whether we actually want advice systems to feel good, or to be right.
ohsecurity··on The 'paperwork flood': How I drowned a bureaucrat before dinnerIt's 2026 and they're still rejecting email for security while asking for 500 page faxes. Hilarious and depressing at the same time.