AI Policy at a Glance What is this about? PMU generally allows the use of AI as a supporting tool. AI may help with learning, writing, research, and structuring work, but it does not replace your own professional, academic, ethical, or clinical responsibility. This policy has been in effect since July 1, 2026 , and applies to all students, lecturers, including external lectures, researchers, and staff members of PMU. (this page is AI generated) 1. You remain responsible Everything you create or edit using AI remains your responsibility. You must review the content, correct errors, and ensure that your work remains academically sound, transparent, and your own independent work. 2. AI may be used in teaching only within the permitted framework Whether and how AI may be used in a course, exam, seminar paper, bachelor’s thesis, master’s thesis, or doctoral dissertation depends on the applicable requirements of the course, supervision, or examination rules. The policy distinguishes between four categories: Category Meaning for students A – No AI use permitted AI may not be used. Only explicitly approved aids are permitted. B – Only linguistic/formal support permitted AI may only be used for spelling, grammar, translation, formatting, or linguistic polishing of your own texts. No content-related support, no generation of solutions, no arguments, no research, and no data analysis. C – AI permitted, but documentation required AI may be used as a supporting tool. Students remain fully responsible and must document the type, purpose, and scope of AI use → “Documentation.” D – AI is part of the assignment AI should or may be used intentionally because it is part of the assignment. The assignment specifies which tools are permitted, how the use of AI must be reviewed and documented, and how AI use will be assessed → “Documentation.” Important: The applicable category must be communicated to students transparently and in writing no later than the beginning of the course or when the assignment is issued. This is especially important for courses with assessments that may be directly affected by AI, such as take-home assignments, unsupervised work, or theses. If no category is specified, Category A applies to supervised exams, while Category B generally applies to written or unsupervised graded work. The categories may also be applied to individual parts of an assignment. For example, Category D may apply to the research/results section of a thesis, while Category C applies to the rest of the thesis. Program directors or the persons authorized by them must communicate this designation clearly, accessibly, and in a timely manner at the beginning of a course or project. This applies only to courses with assessments that are completed without supervision and/or could be created using AI tools. Important: In in-person exams, attention must also be paid to small, portable, and outwardly unobtrusive electronic devices, such as smartwatches, fitness trackers, or miniature earphones, through which information from outside the examination room could be received. 3. AI use must be transparent If you use AI for academic or work-related output, its use must be made transparent, especially if AI has influenced content, wording, analyses, translations, code, images, or other work products. A good disclosure briefly answers the following questions: Question Example Which tool was used? ChatGPT, DeepL, Copilot, Perplexity, etc. What was it used for? Idea generation, structuring, language correction, summarization, translation How strongly did it influence the work? “Linguistic revision only” or “content-related contribution” Was the output reviewed? “All content was reviewed and revised by a person with subject-matter expertise.” As of August 2, 2026 , AI-generated content must be labeled. As a rule of thumb : The more realistic, public-facing, decision-relevant, or trust-building an AI-generated content item is, the more important clear labeling becomes. The information must be clear and distinguishable no later than the first time the content is perceived, for example through a notice such as “AI-generated.” 4. Special caution is required in exams In assessments and exams, AI may only be used if it has been expressly permitted. If AI is not expressly allowed, it is considered an unauthorized aid. 5. Do not enter sensitive data into AI tools You may not enter confidential, personal, or especially sensitive data into AI systems, in particular data relating to patients, students, instructors, staff members, or research participants. Do not enter: names, student ID numbers, email addresses patient data, diagnoses, findings, health data non-anonymized or unpublished research data internal PMU documents unless they have been approved for this purpose third-party texts, images, or works without authorization confidential hospital documents 6. AI results may be wrong AI can generate content that sounds convincing but is incorrect. This applies especially to sources, medical statements, legal questions, numbers, references, and summaries. Therefore: Always check AI responses against reliable sources, course materials, or academic literature. 7. Respect copyright and third-party content You may not simply upload third-party texts, images, course materials, work by other students or colleagues, or protected materials into external AI tools. This may raise copyright or data protection concerns. 8. Deception is prohibited Undisclosed or unauthorized use of AI may be considered a violation of good academic practice or examination rules. Possible consequences may include restrictions on use, additional examinations, documentation requirements, or other measures under study law.