AI Detection in Academia: How Indian Universities Can Safeguard Academic Integrity ^†

Abstract

In recent times, the use of Artificial Intelligence (AI) technologies like ChatGPT-4o within the education sector has become an undisputed fact. AI has transformed the education sector, offering tools that enhance student research and writing. However, the use of AI raises concerns with respect to academic integrity, originality, and authenticity. Indian Universities regulate traditional plagiarism with anti-plagiarism detection systems. Some Indian Universities have also subscribed to AI plagiarism detection systems, but not all of them have subscribed to AI plagiarism detection. The majority of Indian Universities are not sufficiently prepared to identify AI-generated content that is contextually relevant and original, thus bypassing these traditional checks. This study stresses the urgent need for the University Grants Commission (UGC) to introduce advanced AI detection systems across Indian universities. Unlike regular plagiarism checkers, these tools can identify unique writing patterns that suggest AI-generated content. Without such measures, universities risk students using AI to complete assignments and research dishonestly. Through this research, the authors will examine the ethical concerns surrounding AI in academia and highlight the importance of clear guidelines to ensure responsible use. Colleges and universities need proper policies to regulate AI-generated work in student submissions. This study will compare how India and other countries handle AI detection in education, elaborating on the challenges of dealing with AI-generated content. The paper will propose a structured framework for Indian universities, including the use of AI detection tools, ethical guidelines, and awareness programmes to help students use AI responsibly while maintaining academic integrity in a changing educational system.

1. Introduction

Plagiarism in academics has always meant using someone else’s work without giving them credit [1]. To prevent this, universities use plagiarism detection tools like Turnitin September 2024 version and Grammarly 2023 version, which check student papers against a large database to flag copied content. However, the rise in AI writing tools has made academic dishonesty more complex. AI models like ChatGPT can create well-written text that sounds human-made, making it harder to tell if a student has written something themselves. Unlike traditional plagiarism, where students copy and paste from existing sources, AI-generated content does not directly match any published work, allowing it to bypass regular plagiarism checks.

ChatGPT, developed by OpenAI, is a good example of how powerful AI has become in generating written content. It can quickly generate essays, reports, and even research papers based on simple instructions. Because it is easy to access and use, many students turn to it for help with assignments. While it can be a useful tool for learning and improving writing, it also raises concerns when students depend too much on AI instead of conducting their own work.

The following chapter elaborates on the benefits offered by AI tools like ChatGPT in the field of education. While the ethical use of AI accelerates the process of learning and helps when brainstorming ideas, its irresponsible use creates several challenges. Section 3 provides insights into the difficulties of detecting AI-generated content and distinguishing it from human-written work. Section 4 explores the worldwide institutional responses to deter AI plagiarism, followed by an analysis in the Indian context. Section 5 suggests reforms for Indian Universities to address these challenges, focusing on the revision of UGC’s Anti-Plagiarism guidelines. The paper concludes by discussing the future of AI in academia and strategies that can mitigate its impact.

2. Positive Uses of AI in Education

Despite concerns about misuse, AI tools like ChatGPT offer several benefits in educational contexts. These tools can be used ethically to support learning in various ways, such as aiding with proofreading, editing, and generating ideas for research [2]. Additionally, tools like ChatGPT can be employed as tutors to break down or rephrase complex topics until they are clear to students [3].

Furthermore, students can use AI tools to refine their written assignments by providing prompts to make grammatical corrections, improve sentence structures, and enhance the quality of their overall work. ChatGPT is an efficient assistant for research work. It can generate summaries of complex concepts, provide key arguments, and even help students brainstorm ideas for their projects. This saves time and helps students quickly identify gaps in their existing knowledge base and work on potential areas of research.

The use and benefit of ChatGPT is not limited to students; teachers can benefit from AI tools too [4]. It can help them draft lecture notes, summarise lengthy chapters, prepare assignment questions, or even provide innovative ideas to spark discussions in classrooms [5]. It can help educators keep their courses relevant and fresh by suggesting regular updates or interdisciplinary links. Beyond teaching, AI tools are employed for correcting assignments, mostly objective tests. In India, Optical Mark Recognition (OMR) technology and scanners have been in use for a long time to check multiple choice answer sheets by utilising image processing and pattern recognition to automate the grading process (See [6]). By leveraging the use of technology in these repetitive administrative tasks, like making quizzes, correcting assignments, preparing course plans, and assessing exams, educators will have more time to focus on individual student growth and overall improvement of teaching techniques [7].

However, the reliance on AI for content creation raises questions about the authenticity of student submissions and the value of their academic achievements. The overuse or misuse of AI tools hinders the development of critical thinking and analytical skills in students [8]. The effort and thinking required when researching and completing assignments are indispensable for true learning. As students engage in using AI to bypass these learning processes, and more importantly, understanding, they miss out on opportunities to grasp the material, thereby adversely affecting their overall learning experience. Students are often tempted by generative AI’s ability to produce high-quality and error-free text and end up submitting AI-generated work as their own. This compromises the foundational standards of academic integrity and honesty. Moreover, AI tools frequently produce generic content that fails to reflect a student’s personal viewpoints and/or meet specific assignment requirements [9]. It may also include false or made-up information that is derived from unreliable, outdated, or fictional sources, which students put into their work without reviewing or cross-referencing [10]. In the absence of due human oversight and cross-checking, AI always poses a risk of producing misleading data. Thus, the misuse of AI in academia greatly undermines the principles of honest scholarship and academic honesty.

Moreover, the use of AI tools by students also challenges the traditional methods of assessment, which are primarily based on evaluating a student’s original thinking and writing skills. Evaluators are faced with an additional duty of scrutinising the authenticity of work over evaluating the content [11]. Though AI-generated content can be detected using tools like Turnitin or Grammarly, they are not completely foolproof, making it difficult for them to differentiate AI-generated work from human-written work. Thus, the use of AI tools poses several ethical challenges in academic settings, impacting both students and educators equally.

3. Detection and Challenges of AI Plagiarism

As AI-generated content becomes more sophisticated, detecting its use in academic assignments poses a significant challenge. AI-generated content is often well-phrased, free of errors, and grammatically correct, making it difficult to identify as non-human work [12]. Since AI creates new content instead of simply copying it from other sources, traditional plagiarism detection tools, which check for copied text from existing sources, are rendered ineffective.

Most plagiarism software works by comparing and contrasting a student’s work with an existing database of books, articles, and other scholarly documents. If a match is found, it flags the content as copied. However, AI does not copy from a single course; it creates new content based on the algorithmic patterns it has learnt. Thus, even though a write-up is not drafted by the student, it still looks unique, and the source remains unidentifiable. Moreover, even with a slight change, the AI-generated text can be made to look as if it were written by a human. Students can easily rewrite sentences, rearrange paragraphs, use other paraphrasing tools to swap words, or use synonyms, making it even more difficult for detection tools to flag the use of AI in generated content. Furthermore, AI tools like ChatGPT, Bard, and Claude are always updating and improving [13]. These tools now even provide a humanising feature, which essentially humanises the AI-generated work and makes it look and read more human-like and natural. As these tools improve, it becomes even harder to tell whether a piece was created by a student or AI.

Detection tools struggle to keep up with evolving AI tools. This necessitates the development of specialised AI detection tools that can identify the distinct patterns of AI-generated content [14]. For instance, AI detection tools, like Turnitin’s AI Writing Detector and GPTZero, have been curated to identify AI-generated work. These tools sometimes incorrectly label human writing as AI-generated, called false positives, or fail to detect actual AI-generated text, called false negatives. Also, these tools only provide a probability score instead of a clear percentage, making it difficult to hold a student accountable for using AI. Furthermore, AI is always one step ahead. These tools are constantly being improved to make AI-generated text look even more human-like. Consequently, even when universities develop AI detectors, newer AI versions instantly find ways to bypass them. This creates a never-ending challenge for institutions trying to keep pace with AI-assisted plagiarism. Moreover, a new question posed by AI-related plagiarism is whether a student uses AI to generate ideas but rewrites everything in their own words, would that still be plagiarism? Thus, as AI-generated text becomes increasingly indistinguishable from human writing, universities need to continuously update their detection systems, devise clear rules about AI use, and decide when using AI is acceptable and when it crosses the line into dishonesty.

4. Policy and Institutional Responses to AI and ChatGPT

At the global level, a major challenge in using AI in academia was highlighted by a recent study, “Generative AI in Higher Education: A Global Perspective of Institutional Adoption Policies and Guidelines.” [15]. Many universities in wealthier and more developed nations have better AI resources, while students in lower-income regions struggle with limited access to advanced AI tools and language support [16]. Moreover, students who cannot afford premium AI tools or who do not have a stable internet connection may face deprivation if the use of AI is unchecked or unregulated. To address this divide, the study suggests that universities should consider offering free or subsidised AI access and develop AI literacy programmes. Universities in lower-income countries may draw inspiration from institutions like the California Institute of Technology (United States), the University of Manchester (United Kingdom), and the University of Hong Kong, which have taken proactive measures to ensure students have equitable access to Generative AI tools at no additional cost. This is particularly significant for countries like India, which face a significant digital divide and income inequality.

As day-to-day reliance on AI becomes the norm, many students are being accused of misusing AI in their studies, assignments, projects, and exams [17]. Even when AI-generated content is detected, it is not always easy to justify whether a student has used it in an unethical manner. Students may argue that using AI to draft their assignments is similar to obtaining assistance from a tutor or a thesaurus. As per the 2023 Scottish survey figures, more than 600 students were accused of unethical use of AI in academia [18]. For the first time, Scotland also observed university expulsions based on AI-linked cheating and plagiarism. In the famous Massachusetts case, the parents of a high-school student filed a lawsuit against the school authorities for accusing the student of using AI in preparing his project and thus violating the academic integrity guidelines [19]. It was argued on behalf of the student that AI was used only for research and outlining purposes, and not for writing the entire paper. However, U.S. Magistrate Judge Paul Levenson ruled in favour of the school district’s decision to penalise the student on grounds of AI-related cheating. The family’s request to change the grade and remove the cheating record was dismissed. The said project even included fictitious sources, which the school deemed academic dishonesty. While AI-related cheating guidelines were added to the school’s handbook only in the year 2024–2025, and were absent in the relevant year of 2023–2024, the court held that the educator’s judgements should not be second guessed by anyone when it comes to grading and academic discipline. This case raised important questions over academic integrity in light of AI advancements, and the difficulties faced by institutions in scrutinising AI-related plagiarism.

In another case from Ontario, a 16-year-old student was accused of using AI to generate her assignment. Turnitin—a globally popular AI detection software—was used by the institution to generate the AI detection report. Its claim of 98% accuracy in detecting AI-generated or paraphrased content received major public criticism during the pendency of this case. The case highlighted that a claim like this demands special scrutiny because a student faces significant academic drawbacks due to the percentage of the content flagged as AI-generated. Experts also pointed out the high possibility of false negatives and positives in these reports, making it hard for the educational institutions to devise a policy on this issue.

It is clear that the integration of AI tools in academia offers both opportunities and challenges. Though AI has the potential to make the learning process more advanced and support students in their academic endeavours, it also raises pertinent concerns about academic integrity. The traditional methods of plagiarism detection are not sufficient to address the new challenges posed by AI-generated content. Therefore, universities need to take proactive measures to adopt AI detection technologies, establish clear policies, and promote ethical AI use.

5. Why Indian Universities Should Regulate the Usage of AI Detection in Academia?

In India, the authority to frame regulations on minimum standards of education, including teaching, examinations, and research across universities, rests with the University Grants Commission (UGC). It has taken various steps to foster a culture of responsible research, original work, and proper attribution. For instance, the 2018 UGC Anti-Plagiarism Regulations set rules and guidelines to deter plagiarism in academic writing. At the time of submission, it mandated the use of an “appropriate software” to confirm that documents from theses, dissertations, or publications were free from plagiarism [20]. Based on the percentage of similarity detected in a student’s work, traditional plagiarism is categorised into four levels. As shown in

Table 1. Plagiarism Penalty Levels Based on Similarity Index.

Level 0	Similarity up to 10%	No penalty
Level 1	Similarity between 10 and 40%	Revised script within a stipulated time period not exceeding six months.
Level 2	Similarity between 40 and 60%	Debarred from submitting a revised script for a period of one year.
Level 3	Similarity above 60%	Registration for that programme shall be cancelled.

:

Under the current legal framework in India, a student who uses AI to generate most of their assignment can argue that the university lacks a clear policy on AI usage [21]. They may claim that, despite using AI, they retain copyright as the principal author per Indian Copyright laws. While legally valid, such arguments expose gaps in the existing system, raising critical questions about the use of AI in academic work and the ethical challenges of AI detection tools. The following reforms are proposed to bridge this gap and support both students and educators in adapting to AI’s evolution.

5.1. Adopt AI Detection Systems

The UGC should extend its plagiarism guidelines to mandate institutions to adopt advanced tools like Turnitin and iThenticate, which leverage AI for accuracy in detecting plagiarism and AI-generated content. These tools will help ensure students submit original work and do not misuse AI. However, it should also be understood that AI detection tools are not foolproof. The University of Kansas, U.S., in its “Guidance on AI Resources” highlights the limitations of such tools [22]. For instance, while Turnitin claimed 98% accuracy in its AI detection, its margin of error could be as high as ±15 percentage points. This means that a flagged AI-generated score of 55 could actually range anywhere from 40 to 70. In addition, it is stated that these tools were trained on older AI models, which may struggle with detecting modified responses and a mix of human and AI-generated text.

However, instead of becoming caught up in debates over the reliability of individual AI detectors, the UGC could mandate institutions to conduct AI plagiarism checks using at least two different detection tools. It must also be ensured that these tools are regularly updated. This would allow universities to ensure a fairer system for the students.

5.2. Develop Comprehensive Ethical Guidelines

The UGC should create detailed and dedicated rules on how AI can be used in education, prioritising transparency, human oversight, and academic honesty. To give effect to these rules, students should be compelled to duly acknowledge AI-generated content in their submissions. Inspiration may be drawn from Newcastle University, UK, which has introduced a “Student Charter,” which sets out the rights and responsibilities of students in maintaining high standards of academic honesty and integrity. It has also devised an “AI for Learning Canvas” course, which provides practical advice, examples, and resources to help students use generative AI productively and responsibly [23].

The approach adopted by this university showcases a balanced viewpoint toward AI in academia. While recognising AI’s benefits in learning and development, clear ethical guidelines are set for its use. Students are mandated to reference AI sources alongside other materials and critically reflect on how, why, and when AI was used. The UGC should also frame similar ethical guidelines, ensuring mandatory disclosure and acknowledgement of AI-generated work without discouraging its use. The broader goal of framing these guidelines should be to ensure a fair and justified measurement of a student’s progress against the intended learning outcomes while discouraging and/or penalising the use of unethical means.

5.3. Develop a Structured Penalty Mechanism for AI Plagiarism

A structured penalty mechanism for AI plagiarism, similar to that for copyright plagiarism, would bring clarity to existing uncertainties about its use. However, since AI detection tools are still evolving and may not always be accurate, penalties need to be more flexible than those for copyright-related plagiarism. Provided that the use of AI in education is still new, students should be given a fair opportunity to redo their work if suspected of AI plagiarism. If an assignment appears to be AI-generated or highly AI-assisted, teachers may allow resubmissions without any penalty or reduced grades. However, if a student is found to have completely relied on AI with no original effort at all, they may receive a significant grade reduction. Particularly in cases where a student repeatedly misuses or unethically uses AI despite warnings, universities should establish clear disciplinary measures, including reporting such behaviour as academic misconduct, which may lead to debarment or expulsion from the course.

5.4. Implementation of AI Literacy Initiatives

AI literacy programmes are extremely crucial to effectively implement the above reforms. Without a clear understanding of what is acceptable, what is unethical, and what falls into the grey area, both students and educators will keep struggling to navigate the use of AI in education.

The Centre for AI Ethics, Literacy, and Integrity at the University of Calgary has developed a simple and effective literacy programme called the ROBOT (Reliability, Objective, Bias, Ownership, and Type) Test to help students critically assess AI tools before using them [24]. It involves evaluating certain key aspects such as the reliability of available information about the AI–such as whether it comes from a credible source, the purpose of using AI–whether it is meant to inform, persuade, or seek funding, whether it requires human intervention or is fully automated, and the ownership of AI, including whether it is controlled by a private company, the government, or a research group.

The UGC should develop a similar programme with a checklist to help students and educators critically assess AI tools. This would ensure that the users understand what they are dealing with, how AI-generated content may impact academic integrity, and the potential consequences for their academic and professional careers. The UGC already mandates workshops on topics like the National Education Policy (NEP) 2020, Indian Knowledge System, and gender sensitisation to improve teaching and research quality [25]. Introducing a mandatory workshop or literacy programme on the ethical usage of AI for both students and educators would be the most practical step toward ensuring responsible usage of AI in academia.

6. The Way Forward

Firstly, rather than banning the use of AI tools completely, universities should focus on educating students about the ethical use of AI. For instance, using AI to brainstorm ideas, organise thoughts, improve grammar, or improve formatting is not unethical, but using it to generate entire essays is against academic integrity. Teaching students via regular and mandatory sessions on the right way to use AI and educating teachers on the right way to assess the use of AI can help maintain honesty and academic integrity.

Secondly, to counter AI plagiarism, more advanced methods need to be developed to detect AI-generated content. The most common among these is using AI against AI. Using AI models to analyse text and identify patterns unique to AI-generated writing would help detect such work effectively. Some AI models also leave behind subtle digital fingerprints that well-equipped AI tools can detect. Thus, the establishment of clear institutional policies on AI’s academic implications is of the utmost importance.

Thirdly, instead of relying solely on written assignments, universities may also consider giving more weight to in-class writing tests, which involve students undertaking written assignments under due supervision to ensure that they are performing the work themselves. However, this may not be very viable owing to growing e-education models. Universities can devise a step-by-step assessment model, which involves grading students based not only on their final submissions, but on their research process, drafts, revisions, and even their oral understanding of the work.

In conclusion, AI is a powerful tool that offers many benefits, but its use must be balanced with ethical responsibility. Students have two choices: One is unethical, using AI to replace their own thinking, submitting AI-generated work without understanding the subject matter, or relying entirely on AI for research and writing. This creates an adverse impact on their intellectual growth and critical thinking skills. The other choice is ethical usage of AI, wherein AI is used to assist rather than to replace human effort. It may be used to make processes more efficient, generate quick summaries, rectify errors, and help students brainstorm ideas. AI is not seen as a threat but as a tool with affordable and accessible assistive potential. What matters is ensuring that students opt for the latter choice.

Responsible AI use in academia must be guided by a balance between innovation and integrity. While broader UGC regulations on AI usage in academics are awaited in India, it is suggested that universities promptly draft and implement their own guidelines. Clear policies in this regard will not only help students understand what is expected of them in terms of ethical AI usage but also provide teachers with defined parameters for fair evaluation. The ultimate objective is not to resist AI but to shape its role, where technology enhances human intellect rather than replacing it.