A Wuhan-based company was recently ordered to pay 4,000 RMB in damages for infringing on AI-generated artwork created by Wang.
During the holiday break, Putui IP expert Lao Yang researched text-to-image systems, following his earlier analysis: “Who Owns Copyright for AI-Generated Images? Is It Infringement?” As an intellectual property professional, Yang focuses on applications and copyright risks of AI art.
Ownership Models for AI Image Tools
- Platforms & Deployment Methods
- Membership Platforms (e.g., Midjourney): Common globally.
- Local Deployment: Open-source models (e.g., Stable Diffusion, FLUX.1) for computers/mobile. Hardware limitations may apply.
- API Integration: Third-party APIs offer cost efficiency (e.g., fractions of a cent per image). Some base models are free.
- Yang has tested most options and developed an “AI Logo Generator” tool via API.
- Copyright Ownership
- Local Deployment: Highest user ownership.
- API/Platforms:
- Non-members get watermarked images; members remove watermarks.
- Ownership typically shared between creator and platform.
- Complex AI Art: Eligible for copyright registration. High-quality outputs often require dozens of iterations. Unauthorized use constitutes infringement.
Industry Impact
Small businesses previously avoided images due to infringement risks and high penalties. Now, AI-generated art is safely used in:
- Social media posts (WeChat/Weibo)
- Official accounts
- Video thumbnails
- Design/e-commerce applications.