![IL-2023-000007 - [2025] EWHC 2863 (Ch)](https://backend.juristeca.com/files/emisores/logo_O3rEzCI.png){kind=logo}
Local Downloads via GitHub and Hugging Face
Local Downloads via GitHub and Hugging Face
I have dealt in the main Judgment with the potential for this to take place in two ways: using either the Direct Download Method or the Diffusers Method.
The notional assessment of the court requires it to consider the average consumer generating images through each of these mechanisms, as well as what the user understands by way of context from the GitHub and Hugging Face pages themselves, which are not identical for the Models in issue.
- Heading
- Mrs Justice Joanna Smith DBE INTRODUCTION
- FACTUAL BACKGROUND
- PROCEDURAL BACKGROUND
- THE WITNESSES AND EVIDENCE
- LEGAL RESPONSIBILITY FOR STABLE DIFFUSION v1.X
- THE TRADE MARK INFRINGEMENT CLAIM
- The Expert Evidence as to the scope for generation of watermarks*
- Annex 8I
- The Getty Watermark Experiments and Annex 8H
- Re-worded prompts
- Evidence of watermark* generation “in the wild”
- Model v1.x
- Models SD XL and v1.6
- SECTION 10(1) INFRINGEMENT
- Use of a Sign
- Identity of Mark and Sign
- Identity of goods or services
- Getty Images Watermarks*
- SECTION 10(2) INFRINGEMENT
- SECTION 10(3) INFRINGEMENT
- PASSING OFF
- THE SECONDARY INFRINGEMENT CLAIM
- COPYRIGHT SUBSISTENCE AND OWNERSHIP
- THE LICENSING ISSUE
- Sources of law
- The interpretation of written contracts
- REMAINING OUTSTANDING ISSUES
- CONCLUSION
- Appendix A Glossary of Terms
- Appendix B
- I shall address the following scenarios a consumer generating content through a locally downloaded copy of Stable Diffusion v2.0
- Local Downloads via GitHub and Hugging Face
- Stable Diffusion v2.x The Stability GitHub page for v2.x includes the following features
- A General Disclaimer in the following terms
- The “Use-based Restrictions” in Annex A are stated as follows
- The model license is again stated to be subject to a CreativeML Open RAIL++- M License
- DreamStudio (v.1.4 and 2.0)
- Logging into the account, the user is again faced with a large stability.ai logo Conclusions