The role of generative AI in transforming creative industries and copyright law

The intersection of generative artificial intelligence (GenAI) and copyright law represents one of the most profound legal and economic tectonic shifts in modern history. As creative industries—spanning literature, music, visual arts, film, and software development—integrate tools capable of generating complex content from simple text prompts, foundational concepts of authorship, ownership, and fair use are being intensely challenged.

1. The Transformation of Creative Industries

Generative AI has fundamentally altered how content is conceived, produced, and monetized:

• Democratization vs. Devaluation: Tools like Midjourney, Stable Diffusion, OpenAI’s GPT models, and specialized audio generators allow individuals to produce high-grade concept art, music tracks, and written copy instantly. While this lowers the barrier to entry for indie creators, it has also flooded markets with synthetic content, devaluing traditional commercial art, stock photography, and entry-level freelance writing.
• Workflow Integration: Rather than replacing humans entirely, GenAI is heavily embedded as a co-pilot. Studios use AI for storyboarding, rapid prototyping, background music generation, translation, and video game asset creation, drastically shrinking production timelines.
• The Problem of “The Commons” and Scraping: GenAI models rely on training data sets consisting of billions of parameters scraped from the public internet. Creative industries argue this utilizes decades of human-crafted labor without consent or compensation, turning their life’s work into fuel for technology monopolies. DDG Avocats

2. The Core Legal Battlegrounds in Copyright Law

Courts and legislative bodies worldwide are wrestling with several distinct legal pillars:

A. Input: The Legality of AI Training (Text and Data Mining)

The central debate is whether training a machine learning model on copyrighted material constitutes copyright infringement or falls under Fair Use (in the U.S.) or Text and Data Mining (TDM) exceptions (in jurisdictions like the EU).

• Major lawsuits brought by authors, visual artists, and music publishers against AI developers (such as Bartz v. Anthropic and litigations involving Meta and OpenAI) have forced courts to evaluate whether copying data into an AI memory space is “transformative”. AIMultiple
• While some early U.S. rulings leaned toward fair use under specific conditions (noting that models learn underlying patterns rather than storing exact copies), massive class-action settlements (such as the landmark $1.5 billion settlement approval in Bartz v. Anthropic) highlight the immense financial and legal risks tech companies face for unauthorized ingestion. AIMultiple

B. Output: The Ineligibility of Pure AI Generation

Copyright offices globally (including the U.S. Copyright Office and European courts) have solidified the stance that works generated entirely by artificial intelligence without direct human creative intervention are not eligible for copyright protection.

PerspireIP

• Because copyright law is fundamentally anthropocentric—designed to incentivize human authors—machine-generated outputs fall into the public domain.
• However, “AI-assisted” works can secure copyright protection only if a human demonstrates sufficient creative control, selection, arrangement, or modification of the final product. PerspireIP

C. Infringing Outputs (Mimicry and Style)

Another legal frontier involves models generating outputs that closely mimic the distinct style, voice, or likeness of specific human artists (e.g., unauthorized synthetic voice clones or graphic illustrations mimicking a specific illustrator’s portfolio). This touches upon adjacent legal doctrines like unfair competition, right of publicity, and traditional copyright infringement if the output is deemed “substantially similar.”

3. Legislative Evolution and Regulatory Shifts

Recognizing that legacy copyright frameworks are ill-equipped for AI, lawmakers are moving past passive observation toward active regulation:

• The European Union Framework: The European Parliament adopted a landmark resolution addressing GenAI and copyright, setting a strict legislative trajectory. It mandates strict transparency obligations for AI developers regarding training datasets, demands robust, machine-readable opt-out mechanisms for rights holders, and reinforces that commercial deployment without addressing copyright compliance exposes companies to heavy liabilities. LYNX legalЕщё 1
• Licensing Markets and Collective Management: To prevent endless litigation, the industry is shifting toward voluntary collective licensing schemes. Tech companies are increasingly striking multi-million-dollar deals with news conglomerates, stock photo archives, and music publishers to license data legally upfront, bypassing the legal grey area of web-scraping.

Summary Outlook

The relationship between generative AI and copyright law is moving away from the “wild west” era of uninhibited scraping toward a regulated ecosystem. The future of creative industries will likely rely on hybrids of technology and legal licensing—where AI models are built on transparent, opt-in, or compensated data sources, and human creativity is legally protected by strict demarcations between machine automation and true human authorship.