The rapid advancement of Large Language Models (LLMs) has led to their widespread adoption, but also introduced significant security vulnerabilities. To evaluate model robustness, red teaming assessments are conducted to expose security risks and implement countermeasures. Currently, red teaming is performed either manually by experts or automatically using predefined attack datasets. However, manual testing is time-consuming, while existing automated methods lack creativity due to their dependency on fixed datasets. This study proposes an automated, human-independent, and adaptive approach leveraging GFlowNets to identify LLM vulnerabilities by training an attacker model against a specified victim model. This research aims to generate more effective adversarial attacks in English compared to existing benchmarks and introduces a model capable of generating attack inputs in the Turkish language. Blogger's Review: This paper proposes a method for automatically generating attacks against large language models using GFlowNets, which is crucial for evaluating and improving the security of LLMs. By automating red teaming assessments, security vulnerabilities can be more efficiently discovered and addressed, thereby enhancing the robustness and reliability of LLMs.