Devanshi Singh
The man-versus-machine battle between artificial intelligence and human creativity is no longer the
realm of science fiction but is already happening in the real-world courts all over the globe. With
generative AI tools becoming a global sensation, literally powering everything but the kitchen sink,
the legal mechanisms of seeking protection over them struggle to keep up. In the centre of this
international legal drama lies a single vexing question: when machines can replicate human minds,
who owns creativity?
AI developers have enjoyed a certain degree of latitude, at least in the United States, as major tech
firms such as Meta and Anthropic have won significant copyright lawsuits in the country in recent
months. These judgments are a relief to Silicon Valley, yet they also sparks fear among authors, artists
and music producers around the world. The European countries, in sharp contrast, have zeroed in on
a more moderate, creator-friendly approach, especially the United Kingdom and Denmark. Such a
difference of jurisdiction is indicative of more than a mere dichotomy of philosophy; it is a harbinger
of the divided copyright law of the future in the era of AI unless an international concord can be
made.
AMERICAN LENS: PAROCHIAL PERSPECTIVE OF THE AUTHORSHIP
One of the major successes of the AI industry was the decision of a U.S. District Court judge who
rejected copyright claims brought against the company Anthropic(controls the Claude family of AI
models) by a group of authors of AI 1 . The authors claimed that Anthropic had used large amounts
of their copyrighted books without their permission to train its language model, thus infringing their
copyright. The court, however, opined that mere training did not allow a case of infringement, unless
the authors could show that the results of the model were substantially similar to parts of their works
or represented an exact copy of part of their works 2 .
This case marks one of the trends where the U.S. courts support the interests of developers when
relying on the doctrine of fair use that constitutes a legal defense in 17 U.S.C. § 107. The judge
referred to the fact that machine learning could be rightfully considered a transformative usage,
provided the purpose of the AI model differs significantly enough from the initial creative intent 3 .
Soon after the Anthropic triumph, Meta (parent company of Facebook and Instagram) got one too.
At that, the court once again stated that Meta had failed to provide enough evidence that its LLM
models regurgitated copyrighted materials 4 . Nevertheless, there was a cautionary note issued by the
judge, cautioning Meta that subsequent lawsuits can prove to be successful if the outputs of AI are a
close match with the protected works 5 .
The two consecutive rulings, as victorious as they might be to the tech community, highlight how
unsound the creators stand in the American justice system. Up until this point, the latitude given to
1 Anthropic wins key ruling on AI in authors’ copyright lawsuit, The Hindu (June 24, 2025),
https://www.thehindu.com/sci-tech/technology/anthropic-wins-key-ruling-on-ai-in-authors-copyright-
lawsuit/article69734375.ece
2 Id.
3 17 U.S.C. § 107 (2023); see also The Hindu, supra note 1.
4 After Anthropic, Facebook-parent Meta wins AI copyright case but gets warning from judge, Times of
India (June 26, 2025), https://timesofindia.indiatimes.com/technology/tech-news/after-anthorpic-
facebook-parent-meta-wins-ai-copyright-case-but-gets-warning-from-
judge/articleshow/122091931.cms
5 Id.
developers is quite broad, and without explicit evidence of literal copying in AI output, courts seem
to not want to cut innovation short in the name of copyright pedantry.
DENMARK AND THE RIGHT TO BE YOURSELF
In Europe, nevertheless, the tune is quite different. In a recent development, Denmark introduced an
innovative legal system which enables its people to copyright their own faces-a daring attempt to deal
with the threat of AI-generated deepfakes 6 . This extends to the voice, facial structure and other
biometric identifiers, giving people more control on their digital identity. The Danish Justice Ministry
said, AI presents an “existential threat to individual free agency” online, and unless regulated,
malicious parties can create highly convincing simulations to do political, in addition to financial or
social damage 7 . The initiative gives the citizens the power to sue developers or distributors of non-
authorized AI-generated content that resembles people 8 .
In contrast to the U.S., where the AI output is viewed as a broadly transformative or fair use,
Denmark has adopted a two-pronged approach to rogue deepfakes that is more commensurate with
the actual harms of the 21st century: unauthorized deepfakes are considered privacy violations and a
copyright violation simultaneously.
UK’S ARTIST OUTCRY
The United Kingdom is also struggling with increased unrest from the artistic community of society.
To address claims that the AI firms were scraping and reusing content indiscriminately, popular
musicians, authors, and artists initiated protests aimed at pressuring a change in the law 9 . Here is a
songwriter based in the UK complaining: “Our works are not open source simply because they
happen to be online 10 . These artists criticised AI companies for taking advantage of loopholes in UK
copyright legislation, including the new text and data mining exception that became effective in 2014
under Sec 29 of the Copyright, Designs and Patent Act, wherein they enabled AI firms to search web
content without express license 11 .
The outcry has raised concerns amongst the government to engage in consultations over whether the
copyright regime can work in the AI era. Although no changes have been implemented, the
lawmakers in the UK have recognized the stifling impact unlicensed AI training might pose on the
creative economy 12 .
THE GLOBAL SCENARIO
As the U.S. supports developers, Denmark with an eye on the protection of identity and the UK lean
towards stricter norms, we may end up with a disunified international copyright regime. Within this
backdrop, organisations can indulge in jurisdiction shopping, where they can train their artificial
intelligence models within lenient jurisdictions and sell their products worldwide. For example, an AI
model trained in the United States could digest a novel by a British writer without its consent and
create a similar passage and publish it in the EU, leaving the writer with minimal options since the
cross-border enforcement possibilities are limited. This contradiction is troublesome.
CONCLUSION
6 Denmark to allow citizens to copyright themselves to fight AI deepfakes, NDTV (June 22, 2025),
https://www.ndtv.com/world-news/denmark-to-allow-citizens-to-copyright-themselves-to-fight-ai-
deepfakes-8781907
7 Id.
8 Id.
9 UK artists rally against AI copyright loopholes, Regulating AI (June 15, 2025),
https://regulatingai.org/uk-artists-rally-against-ai-copyright-loopholes/.
10 Id.
11 Copyright, Designs and Patents Act 1988, c. 48, §29A (UK).
12 Regulating AI, supra note 9
Legal win by Meta, identity-protective laws in Denmark, and the uproar of artists in the United
Kingdom have made it clear that the reaction to AI and copyright is no longer unified. Artificial
intelligence is getting stronger, and to curb this issue is to not only secure innovations, but to save
humanity as a creative force and human dignity. It is time to redraft the copyright rules in the age of
AI. We need to make sure that machines learn but not replace us without consent, acknowledgement,