Tools to Disrupt the Unethical Scraping of Art: My thoughts on Glaze and Nightshade

 For my anti-scraping tools experiments, I chose a personally meaningful AI-generated image from 2018. Created using an early version of Google's Deep Dream Generator in style transfer mode, the image merges a photograph of my right eye (taken during a rejection episode of my then three-decade-old corneal graft) with a cloud formation.

The creation process involved using the cloud image to establish the training set and then applying this "style" to a photograph of the right side of my face, originally intended as a profile picture. While I was pleased with the full-sized result, the cloud effect became less visible when reduced to smaller dimensions, making it appear simply as an unusual eye image. Though I ultimately didn't use it for its original purpose, this unique composition proved ideal for testing anti-scraping technologies.

This image represents an interesting circular journey—AI-generated art now protected from newer AI generative systems—making it particularly appropriate for this blog post.

Two emerging technologies from research indirectly supported at the University of Chicago offer some hope: Glaze and Nightshade. Though both tools share somewhat technical interfaces and require significant processing time on standard computers, they serve different protective functions:

Glaze embeds inappropriate AI training signals into your images, making it difficult for generative AI to accurately reproduce your artistic style when prompted by your name. My test on an eye/cloud composite image (see above) took approximately three hours on a standard i5 computer. The resulting image appears visually similar to humans but contains embedded neural network weights that should confuse any AI machine learning systems. While this doesn't prevent scraping, it reduces the likelihood of your style being accurately replicated. Which makes "glazing" suited to any artist seeking to protect their unique look.

Nightshade takes a different approach by "poisoning" specific text prompts associated with your work. This tool processes faster (about 30 minutes in my test) and creates images that appear normal to humans but contain corrupted data that could potentially damage AI training datasets. The theory is that widespread adoption might eventually reduce the reliability of AI models trained on scraped content, though this remains speculative. I'm currently testing Nightshade on some of my online photos to gauge any noticeable effects or responses.

Both tools add identifiable suffixes to filenames that might eventually be recognized by scrapers, ironically helping them avoid these protected images. This might in part be what artists want, those internet giants not using or freely sharing our work. Encouraging them to contact us, seek permission and pay to use would of course be better. Additionally, social media platforms typically strip metadata and rename files upon upload, potentially limiting the effectiveness of this possibility.

For maximum protection, artists can apply both techniques sequentially, though the practical impact of this combined approach is still unclear. 

While these technologies represent steps toward better intellectual property protection, they're early solutions in what will likely be an ongoing technological and legal conversation about artists' rights in the AI era.

Is Image Glazing Worth the Hassle?

Protecting our art is becoming crucial in today's AI-driven digital world. But is image glazing the answer? Here's my experience so far.

Cara, a popular alternate platform to Instagram at the moment, has yet to offer built-in glazing. They suggest using Glaze, a separate software. Sounds simple. Not quite.

Setting up Glaze is a bit of a headache:

- It's free, 👍 but requires downloading large zip files 👎

- You need a hefty amount of disk space 👎

- The process feels outdated and tedious 👎

took ~88 minutes to Glaze

The real kicker? It's slow.👎👎👎 We're talking almost an hour and a half per image for my 9x5 submissions on older hardware. Ouch.

tool ~83 minutes to Glaze

You can batch them 👍 but it's one at a time.👎

But here's the biggest issue: you can't tell if it worked 🤞. The image looks the same, and there's no way to verify if it's actually protecting your style.🤞

So, is it worth it? Excuse me if I'm a little sceptical. The process is time-consuming, and we're essentially trusting a black box. 

Can it really save our unique mark-making styles from AI theft?

Protecting Creative Work in the Age of AI Scraping

As creatives in the digital age, we're facing a new challenge: how to protect our work from indiscriminate scraping by AI companies. While tools like Creative Commons licensing have been a go-to solution, their effectiveness against AI data collection is questionable.

Creative Commons: A False Sense of Security?

I've long relied on Creative Commons to share my work while maintaining some control. My license specifies attribution, non-commercial use, and (previously) share-alike terms. However, I'm beginning to question whether this offers real protection against AI scraping.

The Reality of AI Data Collection

Many companies, often hiding behind research organizations, are scraping vast amounts of online data to train AI models. This process often ignores licensing terms and lacks proper attribution or curation.

Changing Tactics

In response, I've updated my blog's license from "share-alike" to "no derivatives," hoping to prevent AI from copying my style. However, the legal landscape around this issue remains unclear, especially in Europe.

New Technological Defences

A promising development is the creation of tools that embed changes in image files. These alterations are invisible to humans but can disrupt AI training, potentially "poisoning" the dataset. Glaze and Nightshade are two such tools, though they're still in development and can be resource-intensive to use.

The Path Forward

Despite these efforts, I'm still uncertain about how to confidently share my work with those who behave ethically while protecting it from misuse. As creatives, we need to stay informed about these issues and continue seeking effective solutions to protect our work in the AI era.

What are your thoughts on protecting creative work in the age of AI? Have you found any effective strategies?

I've prepared this blog post with some good advice and a little rewording from Claude.AI

What is the buzz about Cara?

 

A little while ago I noticed a fair few YouTubers, posting about being time to leave Instagram and most are heading for Cara. I totally understood why they wanted to leave, it wasn’t just the threats of AI  “scraping” up their creations, it also had it origins in changes to the algorithm and move to the more tik tok-ish aspects like short distracting video and capturing attention so the viewer stays on the platform but moves on to the next distraction, then the next. As their attention span gets less they move on down the rabbit hole. Good for Instagram to sell the eyeball time but terrible for a creative trying to find traction, if you don’t make it in the first few minutes and/or the first few viewers, your work will disappear from even your friends, fans or collectors. Ok, good hashtags might let a few interested folk find you, but don’t hold your breath.

The result, you just hear crickets perhaps only a few of your closest friends or colleagues ever get to see your posts now. Worse the posts that you like, the ones that show great accomplishment in your art or photography don’t rate, Whereas an embarrassing photo or unfortunate typos might get shared around. Fun and incentives to post has ebbed away.

The perception that Meta was going to start training their AI systems on user data (hate to tell you but I suspect they already do) set off the alarms in many of the Instagram community. Particularly the original creatives, a sub population I aspire to belong to. Like me I think they are looking for something like Instagram was in the days before it was taken over by Meta.

So what is Cara? Its another platform to post your images (artwork & photos) promoting itself as “by artist for artist”. At the moment is a privately run service that doesn’t incorporate any artificial intelligence images and includes access to Glaze. Which claims to poison the use for neural network training on unauthorized “scraped” images by incorporating hidden codes within the digital version of the image that scramble how the image is represented. I’m still to be convinced this is a good long term solution.

So it is time to have a look, a slow and pondering wander around. I’m sure its going to have a few growing pains.