China’s leading artificial intelligence model companies are beginning to curb the cybersecurity capabilities of their most advanced open-weight models, following a string of high-profile AI hacking incidents involving both US and Chinese companies.