Is your Photo being used to Train Artificial Intelligence? Now you can check

Screenshot of returned results for “music” search on “Have I Been Trained?” website

Lately, there has been a lot of talk about neural network-generated images.  Fake images, videos and more that are AI-generated are causing a stir. These programs analyze photos and content to learn and be trained to recognize certain things. It’s not out of the question that a person might wonder if their own photos are being used in the training sets. There is a brand new site available that is trying to give you that answer.

Spawning AI created image-generation tools for artists, and they just launched a site called “Have I Been Trained?” On the site, you can search through a set of 5.8 billion images that have been used to train popular AI art models. When searching on the site, it will find what it thinks is the closest math based on LAION-5B training data – which is widely used for training AI search terms.

The search tool is interesting to experiment with, as it gives a glimpse into the data that is powering its own processes. As shown in the image above, that screenshot illustrated what showed up when we searched for the word “music”.  You can try to experiment by entering your name, or other keywords that interest you. Be careful – your first search will prompt a warning that basically says a NSFW filter is on – but doesn’t always catch everything.

The tool is not perfect, however. An Ars Technica report revealed that thousands of private medical records were among the photos included within LAION-5B – which raises many ethical and legal concerns. Other searches that were conducted for specific people returned results that were problematic, to say the least.  With these alarming search results, it would be beneficial to be able to remove these records from the data set. The problem with that? It’s not a collection of files you can just delete, but rather a set of URLs that point to images on the web.

In response to this, technologist like Mat Dryhurst and Holly Herndon are spearheading efforts such as Source+, which aims at allowing people to disallow their work or likeness to be used for AI training purposes. It’s early in the process, so only time will tell if these efforts will be effective.

Story via TechCrunch