<![CDATA[The rapid adoption of Generative Artificial Intelligence (GenAI) in education has created significant challenges for conventional assessment practices, particularly in determining whether students' submitted work genuinely represents their competencies. This study aims to develop an AI-Resilient Competency Assessment Model that maintains the validity and authenticity of competency-based assessment in an educational environment where GenAI is readily accessible. The study employed a Research and Development (R&D) approach using Design-Based Research (DBR). The development process consisted of needs analysis, theoretical and conceptual analysis, model development, expert validation, revision, pilot testing, empirical validation, and implementation evaluation. Data were collected through questionnaires, interviews, document analysis, focus group discussions, expert validation, and assessment implementation involving experts, teachers or lecturers, and students. The proposed model integrates eight key dimensions: authenticity, process-based evidence, higher-order thinking, AI transparency, AI literacy, oral defense, reflection, and competency demonstration. The model emphasizes authentic and contextualized tasks, documentation of students' learning processes, transparent disclosure of AI use, evaluation of products and performance, oral verification, and reflective learning. Expert validation and subsequent testing are used to determine the model's validity, reliability, practicality, and effectiveness. The findings indicate that AI-resilient assessment should move beyond dependence on final products and AI-detection technologies toward a multiple-evidence approach to competency judgment. The developed model provides a framework for schools, universities, vocational institutions, and educators to redesign competency-based assessment so that it remains meaningful, authentic, transparent, and relevant in the GenAI era. Future research should examine its implementation across different disciplines, educational levels, institutional contexts, and longitudinal settings.]]>
