Copyright & AI Content
The Ghost-Writer Problem
6 min read
The Ghost-Writer Problem
If a ghost-writer writes your book, you own the copyright. If an AI writes your book — courts are still deciding.
AI models trained on copyrighted books, code, and art raises two questions: was training legal? And who owns what the AI generates? Getty Images sued Stability AI for training on their photos. Writers sued OpenAI for using their books. Meanwhile, the US Copyright Office ruled AI-generated content without human creative input isn't copyrightable. The law is two years behind the technology.
In Plain English
AI copyright raises two issues — was training on copyrighted data legal, and who owns AI-generated outputs? Courts are still deciding both. Practically: AI-generated content you create may not be automatically copyrightable in some jurisdictions, and using AI art commercially carries legal risk.
The Technical Picture
Copyright in AI involves training data rights (fair use vs. licence), output ownership (US Copyright Office requires human authorship), and liability for infringement in generated content (e.g., reproducing training data verbatim). Jurisdictions differ significantly — EU AI Act and India's Copyright Act interpretations are still evolving.
Real-World Examples
- Getty Images v. Stability AI — ongoing lawsuit over training data
- US Copyright Office rejecting copyright for purely AI-generated images
- GitHub Copilot controversy over reproducing GPL-licensed code verbatim
AI-generated content exists in a legal grey zone — understand the risks before using it commercially.