Ideas out of top of my head. 1. Honeypots. Special agents sit online pretending to be children. The abusers found captured immediately, real time. Thi

Ideas out of top of my head. 1. Honeypots. Special agents sit online pretending to be children. The abusers found captured immediately, real time. This requires very quick legal aid channels between countries and organizations. To prevent the abusers from anonymizing (e.g. Tor), strong identification is needed in chats and communication systems - ID card or passport. That's a strong requirement, but better than surveillance, probably. 2. Strong identification of children. At least, children should be reliably identified by chat systems, and all communication with them should require the other party to be strongly identified. Or even full recording of communication, but only when communicating with children. By default chat system assumes user is a child, and to be off the record, user has to prove very reliably he is not a child (and the other side of the communication is a proven grown up as well). 3. AI-generated child abuse videos. Real children not harmed, perverts are satisfied. Although, people enjoying child abuse content deserve punishment anyway, probably. Also, encouraging production of such content increases the danger for normal people (including children) to be exposed to this content. However, the exposure possibility problem exists anyway.