AI tools (e.g. ChatGPT, Copilot, Claude, Gemini) are welcome in your research workflow, provided you follow the rules on allowed/prohibited use, disclosure, and good practice outlined below.

Transparency is a core principle: you must state which generative models you used, for what purpose, and to what extent at every stage of thesis production (drafting, coding, analyzing, revising).

Failing to disclose AI assistance can be treated as attempted deception. Academic integrity demands that you clearly separate your own intellectual contribution from machine support and continuously ensure the technical and factual correctness of all submitted work.

☝️ AI assistance never diminishes your responsibility for accuracy, originality, and ethics.

What You May and May Not Do with AI

For quick reference, the rules follow a traffic‑light system. If an activity is not listed, ask me before you proceed.

✅ Allowed, No Disclosure Needed

Uses described in this section leave no AI‑generated text, code or images in the submitted thesis and therefore do not require a declaration:

• Clarifying concepts or unfamiliar terminology
• Translating literature solely for your own comprehension
• Running standard spell‑checking or grammar suggestions integrated in word processors

☝️ These uses are strictly for private study. Any idea, sentence, or visual that originated from AI and is kept in your thesis — even in altered form - must be disclosed (see Yellow Zone).

🟡 Allowed with Mandatory Disclosure

If any AI output appears — verbatim or paraphrased — in thesis materials, you must list it in the “Declaration of AI Tool Usage” (see below).

Rule of thumb: if any text, code, figure or data from an AI tool survives into your submitted files — even after heavy editing — you must disclose it.

Even when using tools from this zone, the intellectual heavy lifting — problem definition, methodology, critical analysis, data interpretation and final conclusions — must be entirely your own.

☝️ Even if you reword or restructure AI output, the origin must still be acknowledged if it contributed substantively to your thesis.

⛔️ Prohibited

The following uses violate this policy and will be treated as academic misconduct:

1. Submitting AI‑generated text, code, figures or data without disclosure.
2. Letting AI draft substantial thesis sections (≈ two or more consecutive paragraphs) with only superficial edits.
3. Accepting AI‑generated data analyses or statistical results that you have not independently verified.
4. Uploading confidential data, unpublished results or copyrighted material to public AI services without written consent.

☝️ AI may assist at the idea or outline level, but writing of core sections must reflect your own formulation and critical thinking. Generating full paragraphs and making only cosmetic changes is not acceptable.

Intellectual Property & Data Protection

Data Protection

Never share personal data or confidential project information with cloud AI services. Prefer on‑premise solutions or local models when feasible.

Examples of confidential data include identifiable personal information, internal documents from companies, sensitive research datasets, or unpublished findings from collaborative projects.

Copyright

Do not feed course slides, proprietary datasets, or unpublished papers into public AI tools unless you have written approval from the rights holder.

Certainly — here’s the extended § 3 “Disclosure Format” section, with added guidance, examples, and clarification on how to phrase disclosures (including your DeepL Write example):

Disclosure Format

Add a “Declaration of AI Tool Usage” before the bibliography. This declaration should give a transparent and honest account of any AI support used during your thesis work. It must include:

• Tools used: Name of the tool and, if possible, the version (e.g. ChatGPT-4o, GitHub Copilot, September 2024, DeepL Write).
• Purpose: What the tool was used for (e.g. idea generation, text polishing, code suggestions, diagram generation).
• Extent of inclusion: What content from the tool appears in your final submission, and in which parts (e.g. “rewritten introduction paragraph,” “bug fix in listing 4.3,” “figure 5.2 generated using Gemini”).
• Verification: A confirmation that you reviewed and validated the output of the AI tools for correctness, originality, and compliance with ethical and legal standards.

If you used multiple tools for different purposes, group them by task or list them in bullet points. If you used a single tool in multiple sections, indicate the approximate scope (e.g. “used for wording improvements throughout chapters 3 and 4”).

Example: Declaration of AI Tool Usage

I hereby declare that I used the following AI tools during the preparation of this thesis. Their use was limited, transparent, and complies with the AI policy for theses. No AI-generated content beyond what is listed below appears in the submitted version.

Tools Used:

• DeepL Write (as of February 2025) Used to improve grammar, style, and readability in the following sections: – Section 2.3 (“Related Work”) – Section 4.1 (“Evaluation Method”) All revised content was originally written by me and subsequently polished using DeepL Write suggestions. I critically reviewed all changes.

• ChatGPT-4o (OpenAI, February 2025) Used during the planning phase to generate potential headings and subtopics for the introduction and discussion chapters. No AI-generated text was used in the final thesis. Only structural ideas were considered and manually implemented.

• GitHub Copilot (Visual Studio Code, February 2025) Used to assist with code development in chapter 5, especially in suggesting syntax and function scaffolding for the implementation of the core algorithm in Listing 5.2. All code was reviewed, customized, and tested by me. No suggestions were copied verbatim without validation and adaptation.

• Gemini (Google, February 2025) Used to generate a draft version of Figure 6.1 (System Architecture Overview) based on a textual description I provided. The generated diagram served as a visual starting point and was subsequently restructured and refined using draw.io to accurately reflect the actual system components described in the thesis.

I confirm that:

• All AI usage is disclosed here to the best of my knowledge.
• All academic decisions, interpretations, and conclusions reflect my own intellectual contribution.
• I have verified the correctness, originality, and appropriateness of all content submitted.

Consequences of Non‑Compliance

Undisclosed or prohibited AI use constitutes academic misconduct and may result in a failing grade, credit revocation, or disciplinary proceedings in line with university regulations.

Updates

The policy is reviewed annually or whenever major AI or regulatory changes occur. The latest version is always available on this page or on request.

Version: April 2025