Guide to Protecting Art from AI Scraping on Instagram, Google, and the Web
Veronica Winters Painting has released a guide aimed at assisting artists and collectors in safeguarding their digital images against AI scraping. The guide suggests utilizing low-resolution previews, offering partial views, and applying watermarks, along with using cloaking tools such as Glaze and Nightshade. Artists are encouraged to create various display versions, use robots.txt to exclude their work from Gemini training, and apply server-side security measures. For Instagram, it recommends posting contextual lifestyle images and including copyright notices in captions. Copyright remains with the artist unless explicitly transferred, and contracts should ban AI training and data harvesting. Tools like 'Have I Been Trained?' and reverse image searches help identify misuse, and the guide outlines a response plan for infringement, highlighting that strategic decisions can mitigate risks.
Key facts
- Glaze alters microscopic pixel details to disrupt style-mimicking AI; Nightshade turns images into 'poison' data for generative models.
- Google-Extended token allows opting out of content use for Gemini training without affecting search visibility.
- Instagram's Terms of Service grant a non-exclusive, royalty-free, worldwide license to host and use uploaded imagery.
- Copyright attaches automatically upon creation; buying a physical artwork does not transfer copyright.
- Cara is an emerging portfolio platform with built-in anti-AI scraping rules.
- ArtStation and DeviantArt offer 'NoAI' checkboxes that artists must manually opt into.
- The DMCA notice-and-takedown system requires online service providers to remove infringing material upon valid notice.
- Tools like 'Have I Been Trained?' and TinEye help artists check if their work is in AI training datasets.
Entities
Artists
Institutions
- Veronica Winters Painting
- DeviantArt
- ArtStation
- Cara
- Spawning
- LAION
Locations
- United States