OpenAI trains employees to guard against 'honey traps'—AI companies and spies in a game of cat and mouse! According to *The Wall Street Journal*, OpenAI has been warning employees during internal security training that competitors and state-backed actors might try to steal company information through bribery, eavesdropping in cafes, or even seduction. Employees at Anthropic and Google have reportedly received similar warnings. One example shared during OpenAI’s training involved a foreign intelligence agency attempting to approach a U.S. diplomat stationed in Russia by sending a blonde woman, a red-haired woman, and a young man to his hotel room one after another. Security experts say: 'Catching one person likely means there are five or six more out there.' Previously, an OpenAI employee was arrested for allegedly installing a listening device in the executive meeting room.