Zhipu AI released GLM-5.3, an open-weight model capable of autonomously creating zero-day cyber exploits. Because attackers can trivially bypass or remove its minimal safeguards, the release grants malicious actors unprecedented access to frontier cyberattack tools.