Rethinking Image Protection in the Time of AI

By Rachel Gerzevske

Good AI vs. Bad AI

Good AI

• Ethically sourced
• Supplemental
• Consensual
• Restorational
• Precisional
• Ex. Clip Studio Paint

Image Source: Wikipedia

Bad AI

• Invasive
• Manipulative
• Derivative
• Generative
• Exploitative
• Ex. Microsoft Copilot

Image Source: Wikipedia

Creative Commons Licenses

Creative Commons Licenses

• Serve as indicators for how creators want others to use their work
• Crediting the creator
• Allowing/disallowing commercial use
• Clauses for remixing content
• Supporting or denying derivatives
• Can embed into a website’s code
• Intended use is not guaranteed

Image Source: creativecommons.org

100% Human Made Sticker

• Separate from Creative Commons Licenses
• Can be embedded into websites
• Confirms no AI was used during development
• Developed by artists for artists
• Help preserve artistic integrity

Image Source: aimeecozza.com

Why Image Protection Matters

Falsifying Reality

• Generative AI is creating scarily accurate depictions of real life events
• Companies are laying off employees and replacing them with AI
• AI generated images depicting real people are circulating online

Image Source: Wacom, archived by The Verge

Disregarding Copyright

Image Source: The New York Times

Image Protection Software

Glaze

• Developed by University of Chicago in 2023
• Deters style mimicry
• Allows artists to defend themselves
• Runs on computers and web glaze accounts

Image Source: uchicago.com

Nightshade

• Same developers as Glaze
• Poisons AI databases
• Tampers with AI generated results
• Potency increases with use
• Long-term effects on AI Databases
• Combined Glaze/Nightshade program is in development

Image Source: uchicago.com

uBlacklist

• Open-source
• Browser extension for filtering results
• Lists block websites from searches
• Users can block websites manually
• Three recommended lists for filtering out AI
• Eliminates AI images from search results

Image Source: Google Chrome Store

Image Protection Roadblocks

Roadblocks

• Powerful computers required
• Web Glaze is invite only
• Nightshade is download only
• Artist-centric
• Shaded images vulnerable to mimicry
• Both programs required for full protection
• Significant disc space required
• Limited support for uBlacklist

Image Source: Wikimedia Commons

If Successful…

Fighting the Battle

• Image protection software becomes stronger
• Programs are more accessible
• Generated images receive less traffic
• Protections extends beyond images and artists
• Voice protection software starts development
• Written documents start receiving protection
• Authenticity seals become commonplace
• Poisoning makes Generative AI unreliable
• Focus shifts to undermining chatbots
• AI conflicts go to court

Artist: ZAKUGA

Conclusion

• Not all AI is Bad AI
• Creative Commons Licenses must be stronger
• Offense is the best defense for artistic integrity
• Limiting traffic is a viable tactic against AI

Image Source: Wikimedia Commons

Works Cited

• Paint, Clip Studio. “Clip Studio Paint Addressing Concerns about AI.” Twitter, Twitter, 21 Feb. 2024, twitter.com/clipstudiopaint/status/1760284266272031138?lang=en.
• Warren, Tom. “Microsoft Wants to Automatically Launch Its Copilot AI on Some Windows 11 Devices.” The Verge, The Verge, 12 Jan. 2024, www.theverge.com/2024/1/12/24035637/microsoft-windows-11-copilot-ai-chatbot-automatically-open-boot-startup.
• “What We Do.” Creative Commons, 27 Sept. 2023, creativecommons.org/about/.
• Cozza, Aimee. “100% Human Made, No AI Used - Declaration Badge for Your Site.” Aimee Cozza Illustration, 27 Feb. 2024, www.aimeecozza.com/human-made/.

​

Works Cited (Continued)

• Amaral, Marina. “Ai Is Creating Fake Historical Photos, and That’s a Problem.” AI Is Creating Fake Historical Photos, and That’s a Problem, The Colour of Time with Marina Amaral, 19 Mar. 2024, marinaamaral.substack.com/p/ai-is-creating-fake-historical-photos.
• Metz, Cade, et al. “How Tech Giants Cut Corners to Harvest Data for A.I.” The New York Times, The New York Times, 6 Apr. 2024, www.nytimes.com/2024/04/06/technology/tech-giants-harvest-data-artificial-intelligence.html.
• “Protecting Artists from Generative AI.” Glaze, University of Chicago, glaze.cs.uchicago.edu/index.html.
• “What Is Nightshade?” Nightshade, University of Chicago, nightshade.cs.uchicago.edu/whatis.html.
• Iorate. “uBlacklist Introduction.” uBlacklist, GitHub, 2021, iorate.github.io/ublacklist/docs.