Meanwhile, bot protection companies like DataDome have been offering services to thwart scraping for years, but have recently seen a big shift with the rise of generative AI. CEO Benjamin Fabre told WIRED that the company is seeing a surge in customers seeking protection from AI-related scrapers. “His 70% of customers asked him to confirm that DataDome was blocking his ChatGPT and other large language models,” he says.
Companies like DataDome are established, but cater to the needs of large enterprises and charge accordingly. Generally not accessible to individuals. So Kudurru’s arrival is promising, as it offers a free tool aimed at the very general public.
Still, Kudurru is far from a widespread and permanent solution for artists who want to stop AI scraping. Even its creators believe it is a stopgap measure, pending meaningful regulation and legislative action to govern how AI is trained. Most artist advocates don’t think these companies will stop scraping training data voluntarily.
Copyright activist Neil Turkewitz believes this is a “speed bump” for AI generators rather than an industry-wide solution. “I think they’re great. These should be developed and people should use them,” Turkewitz says. “And it’s absolutely important that we don’t see these technological measures as solutions.”
“I applaud the effort to develop tools to help artists,” Crabapple said. “But they end up costing us, and that’s not where it should be. We need to protect our work from being stolen and regurgitated by multibillion-dollar corporations.” We don’t need to play whack-a-mole to get the job done. The only solution to this is legislation.”
larger permanent installation Changes to the way generators are trained are likely to come from the government. It is highly unlikely that a major generative AI company will voluntarily stop his web scraping. Some are trying to improve their detractors by creating an opt-out feature where people who don’t want their work used can request to be removed from future training sets. These measures are seen as half-hearted at best by many artists who want a world where training only happens if you choose to participate.
To make matters worse, rather than settling on a common system, companies are starting to develop their own opt-in protocols one by one, making it take longer for artists to retrieve their work from each generator. (Spawning previously worked on an early opt-out tool for Have I Been Training?, but Meyer says he thinks the fragmentation is “unfortunate.”)
The European Union is the most advanced in developing a legal framework for artists’ consent to AI training. “It’s going incredibly well,” Tourenent said. She is optimistic that the AI Act could be the beginning of the end for free-for-all training. Of course, the rest of the planet also needs to catch up. And her AI method will help artists force models to choose opt-out rather than moving them to opt-in. In other words, we are still a long way from the dream of an opt-in training structure becoming a reality. Meanwhile, we have Kuduru.