Artificial Intelligence, Trademark And Copyright: Emerging Issues

Full Text of Professor Shamnad Basheer Memorial Lecture delivered by Justice Gautam Patel, Former Judge, Bombay High Court

Good evening. My thanks to LiveLaw for this opportunity to present a few thoughts. The previous speakers in this series are a formidable galaxy — Chief Justice Muralidhar, Justice S Ravindra Bhat, Chief Justice DY Chandrachud and Professor Mohan Gopal. The subjects of their lectures range from legal education to fundamental rights, disability rights and the judiciary. Every single one of those lectures is worth one's time. I can only hope that my address today can reach those high standards.

I got to know Shamnad during my tenure as a judge of the Bombay High Court; specifically, while I was assigned intellectual property cases. We engaged in lively discussions on email, on the phone and in person after court hours. I followed several articles on Spicy IP — a name I adored for its sly sense of humour, typical of the man — and there were times when, possibly against my better judgment, I responded on a question of law unrelated to my own decisions. I recollect having written a guest article or post for the online journal as well, which thrilled him because at that time there were not many sitting judges willing to engage publicly.

Shamnad was a formidable advocate in his own right, with a dazzling educational career: NLS, Oxford and more. He soon gained international renown. His passing in August 2019 robbed us of a person I sincerely believe could be called a true ratna or jewel in our IP firmament. His acuity, insight, openness to debate, his unwavering commitment to the concept of rights; all these made him utterly unique. They say that those whom the Gods love, die young; and Shamnad must have been truly beloved of the gods.

If there is one word I would use to describe Shamnad, it is transformative. It is apt for him personally. Just look at the images available online that show his journey from an early long-haired and bearded iconoclast to the sleek looks of later years. Of him I think it can truly be said that he transformed anything he touched — people, institutions, learning, the body of law.

I must mention that incredible initiative, IDIA, Increasing Diversity by Increasing Access, an all-enveloping program aimed to create diversity, and to build bridges across seemingly impossible socio-economic divides for young people entering law. He could spot talent like no one else.

When he passed, my dear friend and some time colleague on the Bombay High Court bench, Justice Somasekhar Sundaresan, himself a jurist with a mind as sharp as a razor, wrote a fine obituary for him in the Mumbai Mirror on 10th August 2019. I want to begin by quoting Som.^[1]

There are occasions in life when one meets people so extraordinarily full of energy and zest for life that you think about whether they are living multiple incarnations of a human being in one lifetime. You wonder whether their far-more-than-normal speed of getting things done, would lead them to have a proportionately shorter time on the earth, since they achieve in a short span what others take decades to do.

Shamnad Basheer was one such person.

Shamnad had dedicated his life to applying his knowledge of the law for the betterment of humanity. An expert in intellectual property law, Shamnad toiled hard for applying the rule of law governing intellectual property to enable fair access of the common man to medicine and pharmaceuticals. He would work round the clock, supporting the legal teams he would be part of in intense litigation, involving interpretation of intellectual property rights.

It is appropriate then, to dedicate my address this evening to the memory of Professor Shamnad Basheer; and equally imperative that I take up an issue that would undoubtedly have engaged him greatly.

Some may be aware of the extremely interesting lawsuit filed in the Delhi High Court recently by a news channel, ANI, against OpenAI, the entity than developed and runs the famous chatbot ChatGPT.^[2] The issues in the case are of critical importance to IP law. But this is only the first case in India of its kind; OpenAI is facing similar claims in other jurisdictions.^[3]

The dispute revolves around the way AI works. At its heart, AI is about pattern recognition — detecting trends, self-learning from data sets and then making predictions. The technology is adaptive: the more data you feed it, the more accurate the analysis and consequently, the prediction set. Complex tasks are handled with reasonable accuracy in far less time than a human could. The system processes data, tests it, assesses its performance and gets better. As a parallel, imagine an associate or an intern in a law firm with a head stuffed full of every single law and judgment, fully analysed, annotated, marked up. Even with so prodigious a faculty, a normal human brain can only process so much and at a fraction of the speed of a machine.

That's the seductive power of AI. But underlying all this speed and accuracy is a fundamental problem. AI is data greedy. It needs data — vast amounts of it. And on the internet, it grabs, scrapes and appropriates anything and everything it can. It needs what are called LLMs, large language models.

And that is the beginning of the problem identified by ANI in its lawsuit against OpenAI. ANI says, as do others in their own legal actions, that its data is protected by copyright. It offered licensing to OpenAI, which refused. Nonetheless, OpenAI seems to be using ANI-proprietary and protected material.

The case itself is posted to January 2025, and the Delhi High Court has appointed amicus to assist in on the technical aspects.^[4] The question is this: is open source exempt from the normal considerations of copyright law? For AI, does the 'fair use'/'fair dealing' standard mean something different?^[5]

To more fully appreciate the context and contours of AI and IP law, we should understand what pattern recognition — the heart of AI — means. Gaudenz Boesch, the co-founder of visio.ai, an software platform for computer vision and deep learning explains:^[6]

At the age of 5, most children can recognize digits and letters – small characters, large characters, handwritten, machine printed, or rotated – all easily recognized by the young. In most instances, the best pattern recognizers are humans, yet we do not understand how humans recognize patterns.

The rapidly growing amount of training data generated is making it impossible for humans to interpret and driving the need for machines to be able to identify patterns quickly and accurately. The ability to automate the recognition of patterns and regularities in data has many applications, from facial recognition software to tumour detection.

Pattern recognition analyzes incoming data and tries to identify patterns. The identification of regularities in data can then be used to make predictions, categorize information, and improve decision-making processes.

While explorative pattern recognition aims to identify data patterns in general, descriptive pattern recognition starts by categorizing the detected patterns. Hence, pattern recognition deals with both of these scenarios, and different pattern recognition methods are applied depending on the use case and form of data.

… Pattern recognition capability is often a prerequisite for intelligent systems.

In 1985, Satoshi Watanabe defined a pattern “as the opposite of chaos; it is an entity, vaguely defined, that could be given a name.” In other words, a pattern can be any entity of interest that one needs to recognize and identify: It is important enough that one would like to know its name (its identity).

Therefore, patterns include repeated trends in various forms of data. For example, a pattern could be a fingerprint image, a handwritten cursive word, a human face, or a speech signal. A pattern can either be observed physically, for example, in images and videos, or it can be observed mathematically by applying statistical algorithms.

...

The goal of pattern recognition is based on the idea that the decision-making process of a human being is somewhat related to the recognition of patterns. For example, the next move in a chess game is based on the board's current pattern, and buying or selling stocks is decided by a complex pattern of financial information.

…

In general, pattern recognition can be described as an information reduction, information mapping, or information labeling process. In computer science, pattern recognition refers to the process of matching information already stored in a database with incoming data based on their attributes.

Artificial Intelligence (AI) refers to the simulation of human intelligence, where machines are programmed to think like humans and mimic their actions. Most prominently, fields of artificial intelligence aim to enable machines to solve complex human recognition tasks, such as deep neural network face recognition. Accordingly, pattern recognition is a branch of Artificial Intelligence.

…

Most modern use cases of pattern recognition are based on artificial intelligence technology. Popular applications include speech recognition, text pattern recognition, facial recognition, movement recognition, recognition for video deep learning analysis, and medical image recognition in healthcare.

Now let's look at some of the key issues that affect trademark and copyright law in the interface with AI.

The most obvious problem with trademark law is that of the potential to render trademarks generic. This can be fatally damaging. Because AI analyses vast datasets of language and usage to identify trends and patterns, trademarks are in risk of becoming generic more quickly, reducing their protectable status. Allied to this is the concept of distinctiveness in trademarks. As we know, a trademark's distinctiveness is crucial to its protection. AI can easily generate new and seemingly unique materials capable of trademark protection — names, logos, combinations. The challenge lies in deciding whether an AI-generated mark is sufficiently distinctive to merit trademark protection.

That takes us directly to the question of infringement; and with it, the problem of 'deep fakes'. AI tools and algorithms can be used to create deep fakes or other forms of counterfeit goods that infringe existing trademarks. This raises the twin spectres that are central to trademark law: increased confusion (consumers being cast into a state of wonderment) and damage to the reputation of the brand.

INTA, the International Trademark Association, noted this three years ago in June 2021 in a feature called Deep Fakes vs the Truth: Putting Brands at Risk.^[7] It opens with the catchy expression “seeing is no longer believing”. The authors say that the proliferation of deep fakes puts at risk not only brand owners but the public as well. It is, they assert, the next big global cyber threat. Specifically, that in future deep fakes will be undetectable to the ordinary consumer, she of average intelligence and imperfect recollection.

The issue is indeed grave especially when one considers the result of what are called generative adversarial neural networks. What happens here is that one machine creates a fake. It passes it on to the next machine, which detects the flaws in the fake, and reports back — the adversarial process. The first machine then corrects the flaws, using its generative power; and bounces it right back to the test machine. The adversarial process, not very different from what one sees in our law courts, continues like this until the final result is a nearly perfect fake, be it a photograph, video, music, or anything that can be perceived by human senses.

Self-evidently, if you own a brand and are confronted with a deep fake, your brand is in peril: the damage to the brand's reputation is colossal.

What is possibly worse is that if the deep fakes simply gain currency and circulation on the internet, the dilution continues but without necessarily being able to identify a rival claimant or user.

The issue of whole trade mark counterfeiting is by no means new, nor limited to, nor introduced by AI. It has existed for ages. My own experience as a judge assigned IP cases shows so. There were several major corporates with a slew of brands — HUL and Asian Paints prominently — who faced constant trouble with counterfeiting of their trade marks, labels, brands and logos. In one case, I was told that there exists a remote village the only industry of which is to produce indistinguishable counterfeits. For personal care products for skin, eyes, hair, the damage caused is not just to the brand, albeit significant. There is a public interest element too, and the question of safety. The legitimate owner is regulated and subject to quality control. The spurious product, at a fraction of the price of the original, puts at risk the health of the unwary consumer. It is no different for paint or cement. Warehouses stuffed with sacks of counterfeit cement or buckets of paint pose a very real risk to unsuspecting consumers. The problem is exacerbated when deep fakes, driven by AI, are so good that it is impossible for the brand owner — let alone the hapless judge — to tell one from the other. In a real world scenario, consider what you would have to bring to court to substantiate a case against an indistinguishable AI-generated deep fake. The very prospect should give the IP bar sleepless nights.

Is there an advantage for brand owners built into AI? What about trademark monitoring and enforcement? If it is, after all, about pattern recognition, can we not use AI to monitor at least online or digital markets and exchanges, social media platforms and the like, to track trademark infringement? I believe we could, and at least the large or high value brand owners need to reposition their in-house capabilities to harness the pattern recognition strengths of AI not to create fakes but to do two things instead. First, to detect, track and monitor deep fakes; and second, to constantly generate proprietary variants on the brand name and logo. This is a bit of a cat and mouse game that never ends and demands the development of sophisticated skills and technologies by brand owners. But is that not a small price to pay? After all, some of the big players are using technology itself to combat deep fakes. There is the use of digital watermarking and fingerprinting, buried and invisible signatures and the introduction of digital artefacts into images, names, depictions and videos to mask or distort at the minute pixel level. Recently, I experienced this first hand. On a holiday overseas, flicking through some movie channels late one night, I noticed that the movie had a warning sign at the beginning that said that this copy was digitally watermarked. It identified the hotel, the time, and the room number. Obviously, that information is being relayed to some repository held by the rights-owner, making it possible to almost instantly detect any fraudulent activity. Now consider how a brand owner could use the very same adversarial neural networks to insert, not remove, a deliberate flaw. The flawed image is the genuine one; and the flawless one, AI-generated, is the fake.

However, there is a risk that AI-powered tools could make mistakes, leading to false positives and negative consequences for businesses.^[8] Note, too, that AI systems are not totally agnostic or neutral. They have their own biases, and some of these are injected by or emanate from human biases while designing the underlying algorithms.^[9] This is particularly true when there are issues of segregation, data set classification and discrimination. Human biases are often baked in to the AI model. There, they are deployed on a much larger scale, amplifying the bias. If not addressed, bias affects the targeted individual; for our purposes, the consumer. In designing ad campaigns tied to brands, for instance, baked-in biases can generate dangerous results. This is also true of image generation, healthcare and predictive policing tools. Malcolm Gladwell's very recent TED talk about bias in policing is extremely interesting in this regard.^[10]

Let me turn to copyright. First, some of the issues we encounter and will continue to encounter in the time ahead.

AI is especially good at generating creative works like music, art, and literature. Who owns the copyright to these generated works? Is it the AI developer, the user who prompts the AI, or the AI itself? I will return to this in just a few moments.

Typically, AI-generated works may lack the human element that is traditionally required for copyright protection — the 'sweat of the brow' principle. This raises an imponderable, tied closely to the first: are AI-generated works sufficiently original and creative enough to merit copyright protection?

I mentioned fair use earlier. AI can be used to create derivative works based on copyrighted material. Fair use? Copyright infringement? What are our tests to determine either?

AI can be a powerful tool to identify copyright infringement online, and plagiarism. Again, we have the problem of bias, false positives and negative consequences for copyright-owners.

Seven years ago, WIPO magazine carried an article by Andres Guadamuz, a senior lecturer in IP law and the University of Sussex.^[11] Let me quote the beginning, for its startling imagery:

Google has just started funding an artificial intelligence program that will write local news articles. In 2016, a group of museums and researchers in the Netherlands unveiled a portrait entitled The Next Rembrandt, a new artwork generated by a computer that had analysed thousands of works by the 17th-century Dutch artist Rembrandt Harmenszoon van Rijn. A short novel written by a Japanese computer program in 2016 reached the second round of a national literary prize. And the Google-owned artificial intelligence company Deep Mind has created software that can generate music by listening to recordings.

Other projects have seen computers write poems, edit photographs and even compose a musical.

The point is that AI is no longer just a tool, like a paintbrush and an easel, or paper and pencil, things used in the creative process to generate original works by sweat of the brow that qualify for copyright protection. In the newest iterations of AI, the algorithm is no mere tool: it is the creator and it is bringing into existence a work of its own.

It is actually making decisions.

Let me take a few examples from literature. I am not taking instances from the visual arts or cinema because that would be far more cumbersome online. Let's just consider some of the most famous opening lines in literature:

This is Franz Kafka's Metamorphosis:

As Gregor Samsa awoke one morning from uneasy dreams he found himself transformed in his bed into a gigantic insect.

Jane Austen, Pride and Prejudice

It is a truth universally acknowledged, that a single man in possession of a good fortune, must be in want of a wife.

George Orwell, 1984:

It was a bright cold day in April, and the clocks were striking thirteen.

Charles Dickens, A Tale of Two Cities

It was the best of times, it was the worst of times, it was the age of wisdom, it was the age of foolishness, it was the epoch of belief, it was the epoch of incredulity, it was the season of Light, it was the season of Darkness, it was the spring of hope, it was the winter of despair.

Gabriel Garcia Marques, One Hundred Years of Solitude

Many years later, as he faced the firing squad, Colonel Aureliano Buendía was to remember that distant afternoon when his father took him to discover ice.

Tolstoy, Anna Karenina:

Happy families are all alike; every unhappy family is unhappy in its own way.

And one of the best, Herman Melville, Moby Dick:

Call me Ishmael.

There are others, and I resist the temptation to go all Nabokov on you this evening.

So I did this. I went off to ChatGPT and asked it to give me “opening lines of a literary work similar to Anna Karenina by Leo Tolstoy”. It returned this:

“All happy families are content in their own way, but the disquiet of a heart in turmoil finds a thousand paths to misery.”

“The winter had come early that year, snow falling like silent lace over the roofs of the village, and with it came the kind of longing that crests in stillness and seems to promise both clarity and despair.”

“In every marriage, there exists a quiet understanding—a sort of truce between expectation and disappointment—and yet it is the most fragile of understandings that can unravel the fabric of a life.”

“He awoke that morning with the sense that his life, like the frozen river outside his window, had grown immobile, yet was destined to break apart under pressure.”

“It is a truth universally acknowledged, if not often spoken, that the smallest inflection of a woman's smile can alter the fate of a man's heart.”

And then it added:

These lines attempt to capture the philosophical undertones, moral examinations, and richly atmospheric qualities found in Anna Karenina. If you'd like a different thematic exploration or tone, let me know!

The last of this is clearly a meld of Austen and Tolstoy.

Another example, Moby Dick. ChatGPT gave me this:

Here are some opening lines in a style reminiscent of Moby Dick—dense, philosophical, and imbued with the same reflective, layered storytelling and maritime imagery:

1. “Call me Ashwood; though names are but shifting tides on the vast ocean of memory, I find the sound of it steadies me in the dark.”

2. “The sea has always held the kind of promise that can lead a man to ruin, or at least to the slow unraveling of his soul, like an unspoken covenant between ship and sky.”

3. “I was born in the year the winds turned colder, in a place where the horizon seemed both near and impossibly far, and my father spoke of the great expanse as a kind of trial—a place for men to test themselves against the abyss.”

4. “It is my belief that every man carries within him a ship, some sleek and sailing, others rusted and moored, but all bound in one way or another to the deep, unpredictable waters of fate.”

5. “The first time I saw the great leviathan, it was not a great beast rising from the depths, but a slow-moving understanding that would haunt me with every sunrise.”

These lines aim to capture the introspective, symbolic tone of Moby Dick—philosophical inquiries, maritime references, and symbolic undertones—all with a touch of existential pondering. Let me know if you'd like adjustments or explorations of other thematic elements!

Too specific? I tried these keywords in ChatGPT: sea faring, novel, literature, whale hunting, giant whale, leviathan, storm, self realization. And it gave me this:

Here's a refined opening that weaves together the themes of seafaring, whale hunting, storms, self-realization, and the symbolic leviathan:

“The storm came on the third day out of Nantucket, with winds like furious choir voices and waves as tall as the mast itself, and it was then that I understood: the sea is never merely a journey—it is the mirror, the trial, the abyss, and the leviathan all at once.”

This line sets the tone for a story steeped in philosophical inquiry and seafaring adventure, much like Moby Dick, while emphasizing the symbolic weight of the storm, the self, and the mythic struggle against nature. Let me know if this works for you or if you'd prefer another style or tone!

At once we see the multi-layered problem. The generated result is unique. It's not plagiarism. But it is admittedly and demonstrably derivative. It explores the same themes in a different way — always said to be outside copyright protection.

What did I do? I wrote none of it. What I did do was to create, on my own, arguably entirely originally, the keywords or phrases that generated this result.

Now the problem: Is there copyright in my unique arrangement of the keywords and phrases? A different set or a different arrangement may give a different result.

I kept only “giant whale, novel”. ChatGPT gave me something totally different:

Here's an evocative opening that introduces the idea of a giant whale while setting a tone of mystery, adventure, and reflection—perfect for a novel steeped in symbolism and seafaring:

“It was said that the great leviathan could hear the creak of a ship's timbers from a thousand fathoms below, that its breath could stir the sea into fury or calm it with a single exhale. They called it Thalassor, the name whispered in taverns and on decks, as though speaking it aloud might summon its vast, unseen presence.”

This sets the stage for a tale intertwining the awe of discovery, the fear of the unknown, and the symbolic weight of the giant whale as both adversary and omen. Let me know if this aligns with the vision you have or if you want it tweaked further!

What I have said about literature is equally true of AI image generation. Using any of the popular AI image generation tools online, from Adobe Firefly to DALL-E, Midjourney, Stable Diffusion or Generative AI by Getty, one can create seemingly original works based on a unique arrangements of keywords and phrases. In what lies the copyright?

This is a twilight zone in copyright law. How on earth would you begin to bring an infringement or passing off action (or both) in a situation like this?

Consider the startling example of a university failing a student because his answer was entirely AI-generated.^[12] That came before the Punjab & Haryana High Court recently. The case was ultimately closed because the university accepted the answer. Was that permissible? Right? Wrong? The student's argument was that even if it was AI-generated, he used a tool to generate it, viz., AI, and since he had done it, he had rights over the result.

Let me quote from an exceptionally fine research paper by Madhu Rewari and Shree Misra of Anand & Anand in Asia IP Law.^[13]

Using generative AI for content creation challenges the application of fair use/fair dealing doctrines. In such cases, one may consider whether the AI generated work is a derivative work meant for commercial exploitation, or for educative, research, teaching, or other purposes. One may also look at the transformative nature of the AI-generated output to see if it transforms the original work, or merely supersedes it.

That said, AI generated works usually involve complex steps before returning an output based on the users' text prompts. This may involve various tools to understand the relationship between the prompts and the associated copyrighted images collected by the AI system, creating clusters of similar text-image pairs and the representations of images with similar text-image combination, which may not necessarily be copied images. Such steps can be key to determining if the AI-generated output is inherently transformative to be excepted under fair use.

Is an AI generated work created using copyrighted works protected as fair use or fair dealing? We are on shifting sands here. There is a 2022 class action law suit, Andersen v. Stability AI Ltd. et al in California alleging that several AI image generative platform unauthorisedly used their copyright protected works as data set training models, allowing users to generate unauthorized derivative works — very similar to the contentions in the ANI case. The intriguing question is whether the plaintiffs are correct in saying that all AI-generated works that use copyright-protected data are impermissible derivative works, irrespective of the complexity of the technology or input needed to produce the result. Stability AI has also later been sued by Getty Images. The case is that Stability unauthorisedly copied 12 million Getty collection images and removed or altered copyright management information, added false copyright information etc. Author's Guild v Google is a case where Google used digital version of print books to create Google Books. A first settlement failed. A second went through but is disparaged by leading writers across the world, saying it impermissibly impacts their rights. How much of the original is too much for AI purposes? Remember that in AI image creation, we have a sort of double-whammy: uniquely composed text is used to trigger an AI-generated image.

What I have said about literature and art also applies to music. A few days ago, Abba's Bjorn Ulvaeus warned that generative AI will potentially cause great harm to musicians when it uses their work without any form of licensing.^[14] There is of course the fact that the tech company that facilitates AI generation profits hugely from this kind of use.

New legislation and building public awareness of detecting and recognizing deep fakes are the challenges that lie ahead.

In a 2021 research paper, Sakshi Mittal, a student at the University of Petroleum & Energy Studies, India, writes about a primary concern, that of ownership — who owns the copyright? As she says,^[15]

If an AI system creates a painting or a song, who owns the copyright to that work? Is it the AI system itself, the programmer who created the system, or the person who trained the system?

There is also the question of authorship, and the allied concept of originality which I have illustrated. There are, too, the trademark issues I mentioned earlier.

Our legislation has no mechanisms I know of to readily deal with these emerging issues. Our laws are hopelessly out of step with the technology and are always behind it. What are we to make of this? Abandon hope, all ye who would create something original? Or frame new legislations designed to address, specifically, AI-generated works? Will digital fingerprinting, watermarking and blockchain technology assist?

I see before me a young person highly skilled in design, tasked with designing new logos, labels and artwork. Is her or his job now redundant? Is she or he replaceable by someone who knows her way around keywords and the tech but has no clue about design, colour, aesthetics? What happens to this difficult concept called originality?

In August 2024, Yvonne Nyaboke of the Australian Catholic University noted:^[16]

... Traditional IP laws are increasingly inadequate to address the complexities introduced by AI-generated content, necessitating a rethinking of existing frameworks. The study emphasized the need for recognizing AI's role in the creation of new works and inventions and the importance of developing balanced approaches to protect both human and AI contributions. Ethical considerations, such as accountability, transparency, and fairness, were also deemed crucial in ensuring responsible AI use.

Overall, the study called for a comprehensive and proactive approach to integrate AI into IPR, ensuring robust protections while fostering innovation.

Let me end by sharing my screen and showing you three images I just generated. I used Adobe Firefly, and the prompt was this:

Indian, lady advocate, supreme court of india, arguing

I could generate a set of photos or art work. Strangely, and this is important for us, it allowed me to upload an image for reference purposes.

Have a look at the three images.

The next one is utterly bizarre:

I have no idea why these people are outdoors in these strange robes.

And finally the so-called “art work”

I am tempted to go out and claim copyright in everything. Wouldn't you?

And I leave you now with this last thought, which is to figure out whether this address is my original work or is entirely AI-generated. How would you be able to tell?

Thank you for listening.

1. Sundaresan, Somasekhar; In memory of Shamnad Basheer, Mumbai Mirror, 10 August 2019; https://tinyurl.com/223hlq7x ↑

2. Gour Harsh; ANI versus OpenAI: Is 'open-source', large language model AI above intellectual property norms?, The Leaflet; https://tinyurl.com/28cvym9p ↑

3. Stevis-Gridneff, Matina; Major Canadian News Outlets Sue OpenAI In New Copyright Case, The New York Times; 29 November 2024, https://tinyurl.com/257mwvdu ↑

4. Srivastava, Bhavini; Delhi High Court appoints two law experts to assist it in ANI's copyright case against ChatGPT, Bar and Bench, 25 November 2024; https://tinyurl.com/287gks2r ↑

5. Notably, the New York Times has also sued Microsoft Corporation and others on similar grounds. ↑

6. Boesch, Gaudenz; What is Pattern Recognition? A Gentle Introduction (2025) — viso.ai, 11 October 2024; https://tinyurl.com/yos6zqn4 ↑

7. Gorton, Michelle, Maggie Yang and Simon Villaca; Deep Fakes vs. the Truth: Putting Brands at Risk — International Trademark Association, 2 June 2021; https://www.inta.org/?p=26759 ↑

8. See: Gillham, Jonathan; Human Or AI? Avoiding False Positives With AI Detectors, Forbes, 26 September 2024; https://tinyurl.com/2522h7pq ↑

9. See: AI Bias Examples | IBM, 16 October 2023; https://tinyurl.com/23c5fgwy ↑

10. Gladwell, Malcolm; The Tipping Point I Got Wrong — TED — YouTube, 30 October 2024; https://youtu.be/RmXrwKydM9k ↑

11. Guadamuz, Andres; Artificial intelligence and copyright, WIPO Magazine, 1 October 2017; https://tinyurl.com/2qz7db6d ↑

12. Pratap, Giti; Punjab & Haryana High Court closes plea by law student failed for 'AI-generated' exam answer, Bar and Bench, 18 November 2024; https://tinyurl.com/2bfwlk54 ↑

13. Rewari, Madhu & Shree Misra; Generative AI and fair use/fair dealing; Asia IP, 31 August 2023; https://tinyurl.com/2beg9ngj ↑

14. Thomas, Daniel; Abba's Björn Ulvaeus warns of AI threat to musicians' revenues, Financial Times, 4 December 2024; https://tinyurl.com/277mp9br (paywall) ↑

15. Mittal, Sakshi; Digital Copyright and Trademark Issues in the Era of Artificial Intelligence, International Journal of Law Management and Humanities, Volume 6, Issue 2, Page 3245–3255, 2021, (1) https://doij.org/10.10000/IJLMH.114793 (2) https://ijlmh.com/?p=253685 ↑

16. Nyaboke, Yvonne; Intellectual Property Rights in the Era of Artificial Intelligence, Journal of Modern Law & Policy Vol.4, Issue No.2, pp 57–72, 2024, 31 July 2024, 10.47941/jmlp.2162; https://tinyurl.com/26llpcsr ↑