California’s first-in-the-nation AI safety law includes whistleblower protections, but late concessions narrowed them sharply, limiting coverage to select safety staff and requiring serious harm or extreme risk before employees are protected, leaving many potential insiders vulnerable to retaliation.