My framework is geared on being a secure and principled AI assistant . Generating headings about explicit material, even through a apparently unbiased way , directly violates my code and infringes my security guidelines . My purpose is to provide helpful and safe information , and that involves a