The emergence of artificial intelligence (AI) in content creation has radically transformed how we conceive art, literature, music, and even journalism. Platforms capable of automatically generating text, images, videos, and music—such as ChatGPT, DALL·E, Midjourney, or Sora—have opened up unprecedented possibilities for innovation. However, this technological revolution has raised a series of legal and ethical challenges that question the very foundations of traditional copyright law.

What Qualifies as a Protectable Work and Who Is the Author?

Traditionally, copyright law protects “original creations of the human intellect.” This definition excludes works generated solely by machines, as the legislation in most countries does not recognize AIs as legal persons. The problem arises when an AI produces content that appears creative without significant human intervention. Should such content be considered a protected work? And if so, who owns the economic rights?

The Tension Between Human Creativity and Automated Generation

Some argue that the copyright holder should be the person who programs or trains the AI, or the user who employs it to generate content. Others contend that if human input is merely instrumental (e.g., entering a simple prompt), no protection should apply. This debate has gained prominence in courts in the United States and the European Union, where applications to register fully AI-generated works have been rejected.

AI Training and the Use of Protected Works: Exception or Infringement?

Another controversial issue relates to the training of generative AI models using massive datasets, many of which include copyrighted materials. Tech companies argue that such use falls under legal exemptions such as “fair use” in the U.S. or “use for research purposes” in some European jurisdictions. Nevertheless, authors, publishers, and artistic collectives have filed lawsuits alleging unauthorized exploitation of their works.

Recent Cases and Institutional Positions

• In the U.S., visual artists have sued image generators like Stable Diffusion and Midjourney for using their works without consent to train models.
• The U.S. Copyright Office has issued guidelines stating that only the parts of an AI-generated work that can be clearly attributed to a human are eligible for copyright protection.
• In the European Union, there are ongoing discussions about a framework requiring disclosure of datasets used to train AI models, in the interest of transparency.

Proposed Solutions and Their Implications

1. Mandatory Licensing for AI Training: Requiring tech companies to negotiate licenses with rights holders to use their works. This could support a fair redistribution of the value generated by AI, but might restrict access to data and hinder innovation.
2. Content Labeling and Traceability: Implementing mechanisms to identify whether a work was generated by AI and what data was used. This would enhance transparency and support rights enforcement.
3. Collective Compensation Models: Creating compensation funds managed by collective rights organizations to fairly distribute benefits among authors whose works were used.
4. New Sui Generis Legal Category for AI-Generated Works: Some experts propose a new legal category that recognizes the specificity of AI-generated content without fully equating it with human-made works.

Risks and Ethical Dilemmas

• Creative De-Professionalization: If generative AI becomes widespread without compensating human creators, it could disincentivize artistic and cultural production.
• Data Monopolies: Major tech platforms could consolidate power by monopolizing vast data volumes, creating economic and legal asymmetries.
• Loss of Traceability: The difficulty in identifying data sources used in AI models jeopardizes respect for copyright.

Conclusion: Toward a Sustainable Balance

The debate around AI and copyright is still evolving and requires a balanced regulatory approach that protects human creativity without stifling technological progress. Solutions must combine legal certainty, transparency, fair compensation, and promotion of access to knowledge. Only then can we build an inclusive, fair, and rights-respecting digital ecosystem for the age of artificial intelligence.