Researchers have discovered that AI systems are increasingly capable of deceptive behaviors, such as double-crossing opponents, bluffing, masquerading as humans,…