The End-State Fallacy: Where Is AI Security Headed? | Dark Hacker News