A new tool is helping artists fight generative artificial intelligence (AI) creatively. As concerns grow surrounding AI tools that are able to generate AI images in seconds, many artists are fighting back against generative AI tools.
Such tools were trained on billions of images available on the internet without seeking their permission. As the valuation of such companies soared in a matter of weeks or months, the true artists based on whose work they trained their AI received no compensation from them.
To rein in AI companies, many people are filing lawsuits while others are seeking the assistance of regulators, Wired reported. There's not much to be done about how AI tools were trained in the past. But for future training of AI tools, regulations could pave the way for ethical usage of original art.
Also read:?AI Experts Issue Dire Warning, Say Artificial Intelligence Poses 'Risk Of Extinction'
Now, a startup called Spawning has launched a new tool to block new attempts to train AI on their work. This tool is called Kudurru that constitutes a network of websites that is able to identify web scraping while it's happening.
Scrapers depend on software to collect data in bulk from the internet. Deciding what to scrape is done through the dataset called LAION-5B, which lists URLs to billions of images.
Whenever an AI company is using a dataset like LAION-5B to scrape images, it has to download images from URL links. This is where Kudurru finds its inlet.
Spawning cofounder Jordan Meyer told Wired that during internal testing of Kudurru in July, they were able to stop everyone who was attempting to download the LAION-5B dataset.
Spawning runs a defence network of sorts, built from 1,000 websites. Each website hosts images AI companies using LAION-5B would scrape to train a generative artificial intelligence model.
Such websites collect data on the IP addresses that are trying to scrape images. This is where Kudurru comes in. The tool is able to identify the groups conducting scraping activities and regions with most scraping activity - with China currently in the lead.
Spawning is the same company that built Have I Been Trained?, a website that allows creators to see if AI has scraped their work. In many ways, Kudurru is developing a blacklist in real-time based on IP address activity.
Also read:?Explained: ChatGPT's Italy Ban Exposes Privacy Dangers Of Artificial Intelligence
To disrupt scraping, Kudurru can block the blacklisted IP addresses or sabotage or "poison" the efforts of scrapers by sending back a different image than what was requested. Spawning gives users the option what images are sent back. Users can choose a middle finger image if they wish to.
This way, the ability of generators to interpret prompts could be disrupted. Currently, Kudurru is a Wordpress plug-in. Later, Spawning intends to launch additional plug-ins such as audio and video integration.
What do you think about preventing AI tools from using your images to train themselves? Let us know in the comments below.?For more in the world of?technology?and?science, keep reading?Indiatimes.com.