Nightshade and Glaze: How artists are poisoning the AI well

Do you all remember glitter traps? 

Once it became possible to record and share porch pirates taking poisoned bait, watching theives get their just desserts became a staple of local news, email chains, and Reddit superthreads alike. It was like if the term ‘revenge porn’ meant something cool instead.

And we love to see it! If you didn’t want a face full of glitter and fartspray—well, you should have been a better person and not stolen from anyone. Duh. 

Now in the age of generative AI making it easier to starve professional artists, the creatives of the world are fighting back in a similar way. 

Nightshade and Glaze, both lead by Shawn Shan, are pixel-pilferer poison—representing active attacks and selective shielding respectively when it comes to cyber swagger jacking. In the co-companys’ own words: 

“Glaze is a defensive tool that individual artists can use to protect themselves against style mimicry attacks, while Nightshade is an offensive tool that artists can use as a group to disrupt models that scrape their images without consent (thus protecting all artists against these models). Glaze should be used on every piece of artwork artists post online to protect themselves, while Nightshade is an entirely optional feature that can be used to deter unscrupulous model trainers. Artists who post their own art online should ideally have both Glaze AND Nightshade applied to their artwork. “

In its first five days of launching, 250,000+ people zoomed to Nightshade to get their intellectual property protected. Predictably, people too cheap and unskilled to get their art the right way had fits at the very idea that there was something people could actually do to keep their hands off, and as such, it makes one wonder what’s next.  

Normally what I’d say is we can expect is some sort of back and forth evolution, like acacia trees and giraffes. Trees got taller to avoid being eaten, so giraffes got taller to eat them. Trees evolved giant, stabby thorns, so the giraffes selected for tougher lips and more flexible tongues. The trees learned to communicate a warning to other trees as they got munched to push up bitter tannins, and the giraffes learned to start eating upwind. 

Even the folks behind the tools admit the possibility, with Glaze saying: 

“Systems like Glaze face an inherent challenge of being future-proof (Radiya et al). It is always possible for techniques we use today to be overcome by a future algorithm, possibly rendering previously protected art vulnerable. Thus Glaze is not a panacea, but a necessary first step towards artist-centric protection tools to resist AI mimicry.”

Nightshade’s About page admits the same: 

 “As with any security attack or defense, Nightshade is unlikely to stay future proof over long periods of time. But as an attack, Nightshade can easily evolve to continue to keep pace with any potential countermeasures/defenses.”

But I see a different scenario playing out. 

We’ve taken a look at how Reddit offered its userbase to Google to use as AI training fodder. Similarly, in November of 2022, art hosting/social media site (/teen April hangout) DeviantArt announced that works published there would feed DreamAI, their own proprietary image amalgamator. It’s worth noting that after scathing criticism from an already disillusioned userbase saw them walk back the automatic opt-in, but the precedence for hosts serving up their users is well established. And rather than pay someone to develop code to crack digital watermarks, naturally more businesses will see simply siding with (See Also : Paying a Subscription Fee to) web giants to gain easy access as the smarter move. 

For right now, these developments are a huge win for artists’ protection. But there’s always a bigger fish. Whether your business will be a remora or a free swimmer is your call.