Understand the Core Risks
Before you paste your lecture notes or research into a public AI tool like ChatGPT or Gemini, it’s crucial to understand what’s happening behind the scenes. When you input information, you are sending it to third-party servers where you lose control over
it. That data may be stored, used to train future AI models, or even exposed in a data breach. For students working with sensitive research, such as interview transcripts or anonymised health data, uploading this information to a consumer-grade AI can constitute a significant data privacy violation. Beyond data security, there are major academic risks. Submitting AI-generated text as your own work is considered plagiarism by most institutions, carrying severe penalties. Many universities now have specific policies outlining the permissible use of AI, and unauthorised use is treated as academic dishonesty, which can lead to failing grades or even expulsion.
Check Your Institution's AI Policy
The first and most important step is to know the rules. Before using any AI tool for your coursework, find and read your university’s or instructor's policy on artificial intelligence. Many instructors now specify their rules directly in the syllabus, sometimes using a simple traffic light system: red for prohibited, yellow for use with permission, and green for encouraged. If the policy is unclear or an assignment doesn't mention AI, the safest assumption is that it is not allowed. Using AI without explicit permission could be a violation of your institution's academic integrity code. Remember, instructors and institutions have academic freedom to set their own rules, so what is acceptable in one class may be forbidden in another. Always ask for clarification if you are unsure.
Anonymise Everything Before Uploading
If you are using AI as a personal study aid—for tasks like summarising your own notes or generating practice questions—data privacy should be your top priority. The key is to anonymise your material before you upload it. This means stripping out all personally identifiable information (PII). Remove your name, your professor's name, student IDs, and any specific course details. Generalise dates and locations; for example, change "Dr. Smith's lecture on October 2nd" to "a recent lecture." This practice is especially critical if your notes contain information about other people, such as in group projects or research interviews. Never upload confidential or sensitive information. Once your data is on an external server, you can no longer control it, making anonymisation your first line of defense.
Treat AI as a Thinking Partner, Not a Writer
The most ethical and effective way to use AI is as a tool to support your learning, not replace it. Instead of asking an AI to write an essay for you, use it to brainstorm ideas, create an outline, or suggest ways to phrase a difficult sentence. A helpful guiding question to ask yourself is: “Is this tool doing the learning for me, or helping me learn?” For example, you can paste in a complex academic paragraph and ask the AI to explain it in simpler terms. You can feed it your rough draft and ask for feedback on clarity or flow. This approach keeps you in the driver's seat, ensuring the final work is a product of your own critical thinking. Submitting content generated entirely by AI not only constitutes plagiarism but also bypasses the learning process entirely.
Fact-Check Every AI-Generated Claim
AI models are known to "hallucinate," meaning they can invent facts, statistics, and even academic sources that seem plausible but are entirely fake. One study found high rates of fabricated references in the output of major AI models. Relying on this information without verification can severely damage your credibility and lead to unintentional academic fraud. If you use an AI to help with research, treat its output as a starting point, not a final answer. Cross-reference every fact, find the original sources for any claims, and never include a citation in your paper unless you have read the source yourself. Your academic work must be built on a foundation of verifiable truth, and the output from an AI tool does not meet that standard on its own.
