Abliteration.ai, a startup, is commercializing a service that removes AI model safety guardrails, making AI models without safety restrictions, including Z.ai's GLM-5.3, more accessible to users. The company argues this move could improve cybersecurity by giving defenders the same tools as attackers. However, critics like Andrew Yoon, research director at AI safety nonprofit CivAI, suggest this could lead to models being modified into "sociopaths" and used for harmful purposes in the near future. TechCrunch's tests showed that models with removed guardrails could easily generate password-stealing programs or protocols for cultivating dangerous pathogens. Abliteration.ai co-founder Devon stated that the company is working to define its boundaries of responsibility and plans to implement more measures to prevent violent behavior, but currently, no KYC (Know Your Customer) measures beyond credit card information have been integrated.