The deep integration of AI and education relies on the non-intrusive collection and fusion of educational data, and while such a data-intensive research paradigm can promote the development and reform of intelligent education, the collection and analysis of educational data also brings risks of privacy infringement and unethical applications. Data destruction, as a key link in the life-cycle management of personal education data, is also an important means of protecting the ‘’right to be forgotten’’ of education participants. However, the lack of reliable mechanisms for implementing proper data destruction seriously constrains the healthy and sustainable development of intelligent education. This study introduces the concept of Machine Unlearning from computer science fields into the education field, outlines a technical model of intelligent unlearning of educational data, and explains the practical process by which educational data can be intelligently unlearned, taking facial data and graph structure data commonly as examples. The results demonstrate that the intelligent unlearning of educational data can be achieved efficiently, without retraining the original model, ensuring that the unlearned data remains unrecognizable and thus protecting education participants’ “right to be forgotten.” This study provides a valuable reference for optimizing the whole life cycle management of educational data and establishing more robust trust in the digital transformation of education with AI.