Slow fog has two main applications in the field of AI research: 1 Improve the efficiency of our existing work (Hacking driven); 2. Addressing the demand for security defense in the business scenarios of customers in the cryptocurrency field, such as AI/MCP, is more urgent as once security issues arise, funds may be stolen. By the way, if our research output on AI security is useful for non cryptocurrency fields, we would also be very happy