From academic integrity to credential integrity: What does management education certify once artificial intelligence can produce the work?

The debate on generative and agentic artificial intelligence (AI) in management education has centered on whether students may use AI, yet the larger question is what student work can still prove. Once AI can draft and polish a case analysis, an essay, or slides, polished work stops being strong evidence of what a student can do. This article calls the resulting problem credential integrity (i.e., whether a business school can defend what its degree certifies when AI may have produced the graded work). Policies on detection, disclosure, and originality cannot solve this problem, given that all three ask who produced the work and leave intact the assumption that good work proves a capable student. In response, this article draws on assessment scholarship, policy research, and validity theory to propose a verified-judgment architecture in which high-stakes certification rests on unaided command of core concepts and methods, reasoning defended when questioned, positions reconsidered when challenged, learning transferred to changed conditions, and accountable decision making. The architecture is translated into program-level responsibilities for school leaders and redesigned assessments for teachers, neither of which requires a ban on AI or the false belief that an AI-free submission proves learning. A reality check of 84 business schools shows how far practice is from this standard, given that schools describe AI access, curriculum integration, and faculty development more fully than any proof of what the degree certifies and only one school in seven reports a safeguard (e.g., a defended capstone or work kept free of AI).