Why Keep Talking About AI?
A Legal Perspective on the Risks for Authors
By Elisa Bourdier
The more we talk about artificial intelligence (AI), the more complex the legal landscape becomes—particularly for authors and artists. Initially, concerns centred around AI-generated content: would machine-made outputs compete unfairly with human creativity? But the more pressing legal issue lies deeper—in how AI systems learn.
AI models, especially those built using deep learning, require enormous datasets. These datasets are frequently composed of existing works—texts, images, music—scraped from the internet. This practice raises a fundamental legal concern: the unauthorised reproduction and processing of protected works. In copyright terms, this constitutes potential infringement.
As The Guardian recently reported, the UK is actively considering reforms to its copyright regime to accommodate AI development. But the critical question remains: how can lawmakers ensure innovation does not come at the expense of creators’ rights?
To better understand what’s at stake, it is helpful to compare the UK’s approach with that of the European Union, which has recently adopted a comprehensive legal framework for AI.
The European Union’s Approach: The AI Act and Author Protection
The EU—shaped by strong author-protection traditions in countries like France—follows a civil law model that places the author at the heart of its legal framework. The EU’s use of the term “author’s rights” (droits d’auteur) instead of “copyright” underscores this distinction.
The AI Act, adopted in early 2025, is a landmark piece of legislation that categorises AI systems by risk level—minimal to unacceptable—and imposes obligations accordingly. Of particular relevance is Title IV, which targets high-risk AI systems, including those that could affect fundamental rights such as privacy, equality, and intellectual property.
Implementation is phased:
- The ban on AI systems posing “unacceptable risk” took effect on 2 February 2025.
- Codes of practice are due within nine months.
- Transparency obligations for general-purpose AI systems (such as ChatGPT) will apply within 12 months.
- High-risk AI systems will have 36 months to comply.
One of the most significant advancements is the requirement for transparency. Earlier drafts of the legislation merely offered authors an “opt-out” mechanism to prevent use of their works in training datasets. However, without access to or visibility into the datasets used, such opt-outs were practically unenforceable. The final version of the AI Act changes this. It now requires developers to disclose data sources, enabling authors to identify unauthorised use and assert their rights.
Importantly, the EU does not legislate in a vacuum. Through what legal scholars term the “Brussels Effect”, EU regulations often become de facto global standards due to the Union’s market size and regulatory influence. Much like the GDPR reshaped international data protection norms, the EU hopes its AI Act will nudge other jurisdictions towards similar safeguards. Canada is already drafting legislation inspired by it.
The United Kingdom’s Position: Striking a Delicate Balance
In contrast, the UK is pursuing a more innovation-friendly stance—potentially at the expense of authors. As The Guardian reported, artists including Elton John have warned that proposed copyright reforms risk “riding roughshod over the traditional copyright laws that protect artists’ livelihoods.” This concern is widely shared across the creative industries.
The Government faces a delicate balancing act: attracting AI investment while upholding creators’ rights. If artists and authors begin to lose trust in the protections afforded to them, they may withdraw from the creative economy entirely. After all, the incentive to create is often anchored in the assurance that the law recognises and defends one’s contribution.
This concern is further amplified by admissions from companies like Meta, which acknowledged that their AI systems were trained using vast quantities of data—including private content—without obtaining consent. If such practices remain unchecked, it could amount to an expropriation of creative works without compensation.
Perhaps the UK would benefit from drawing inspiration from the EU model, which seeks to balance technological progress with fundamental rights. A framework rooted in transparency, accountability, and enforceability could serve both innovation and the creative sector.
Conclusion: Watching the Future Unfold
As the UK Parliament prepares to vote on its new Copyright Bill, all eyes are on Westminster. Will the government side with technology developers at the cost of creative professionals, or will it adopt a more balanced framework that respects both innovation and authorship?
Wednesday’s vote may well offer an early answer. In the meantime, legal professionals, creators, and policymakers must remain actively engaged in shaping an AI-driven future that honours the contributions—and the rights—of authors and artists.
Elisa Bourdier
Elisa is a French lawyer gaining valuable legal experience at Lawdit Solicitors.
She can be reached at: elisa@trademarkroom.com or via info@lawdit.co.uk
