STAR Indonesian Prompt Injection Dataset

The <i>STAR Indonesian Prompt Injection Dataset</i> is a curated collection of Indonesian-language prompts designed to evaluate and improve the robustness of AI models against prompt injection attacks. The dataset includes a diverse set of benign, adversarial, and mixed-intent inputs that reflect real-world interactions in Indonesian. It aims to support research and development in AI safety, model alignment, and secure prompt engineering. This dataset can be used for benchmarking model vulnerability, training defensive mechanisms, and studying language-specific injection patterns in Indonesian.