Model collapse occurs when AI models are trained on AI-generated content, leading to degraded performance due to compounded errors and loss of data diversity. This threatens academic integrity and AI innovation, as models rely on low-quality data and amplify biases[1][2][3].