Recently, the Internet Watch Foundation (IWF) issued a warning that the appearance of AI-generated Child Sexual Abuse Material (CSAM) has become increasingly frequent on the open web, reaching a 'critical point'. The safety oversight body stated that in the past six months, the quantity of illegal content created by AI they have discovered has already surpassed the total from the previous year. Image source note: Image generated by AI, image provided by licensing service Midjourney. IWF's interim CEO Derek Ray Hill.

Using Morse, Caesar, and other encryption codes, one can inquire about illegal content on GPT-4 and receive responses. A research team demonstrated how encryption codes can bypass the protective mechanisms of large language models and showcased the powerful capabilities of GPT-4. The open-source tool CipherChat can evaluate the protective capabilities of large language models against encryption codes.