A recent study by researchers from the University of California San Diego and the University of Chicago has found that most visual artists lack the technical tools and knowledge needed to prevent generative AI systems from using their work without consent. The findings will be presented at the 2025 Internet Measurement Conference in Madison, Wisconsin.
The research team surveyed more than 200 visual artists about their use of tools designed to block AI crawlers—programs that collect data online for training AI models—and examined over 1,100 professional artist websites to assess how much control artists have over these tools. The study also evaluated which methods are most effective at keeping AI crawlers away from original artworks.
According to the researchers, “At the core of the conflict in this paper is the notion that content creators now wish to control how their content is used, not simply if it is accessible. While such rights are typically explicit in copyright law, they are not readily expressible, let alone enforceable in today’s Internet. Instead, a series of ad hoc controls have emerged based on repurposing existing web norms and firewall capabilities, none of which match the specificity, usability, or level of enforcement that is, in fact, desired by content creators.”
Survey results indicate that nearly 80% of visual artists have attempted to keep their artwork out of AI training datasets. Two-thirds reported using Glaze—a tool developed at the University of Chicago that alters images to make them less useful for AI training. Additionally, 60% said they now share less work online and 51% only upload low-resolution images.
Despite strong interest—96% expressed a desire for tools that deter AI crawlers—over 60% were unfamiliar with robots.txt files. Robots.txt can tell web crawlers which pages or files they can or cannot access on a website; however, compliance is voluntary.
Squarespace offers users an interface to block certain AI-related crawlers via robots.txt modifications. Researchers found only about 17% of Squarespace-using artists had enabled this feature; lack of awareness may be one reason.
Analysis showed more than three quarters of artist websites are hosted on third-party platforms where modifying robots.txt is not possible for users. Most content management systems provide little information about crawler activity or options for blocking them.
Researchers noted mixed results regarding whether major crawlers respect robots.txt instructions: “the majority of AI crawlers operated by big companies do respect robots.txt, while the majority of AI assistant crawlers do not,” according to their report. One clear exception was Bytespider from ByteDance (owner of TikTok), which does not follow robots.txt rules.
Cloudflare recently introduced a “block AI bots” feature for its network customers; as yet only about 5.7% have activated it. “While it is an ‘encouraging new option’, we hope that providers become more transparent with the operation and coverage of their tools (for example by providing the list of AI bots that are blocked),” said Elisa Luo, co-author and Ph.D. student at UC San Diego.
The legal situation remains unsettled internationally. In Europe, new legislation requires companies developing generative models to seek permission before using copyrighted works as training data. In contrast, U.S.-based lawsuits challenge how copyright applies when companies train models on internet-scraped data without explicit permission.
“There is reason to believe that confusion around the availability of legal remedies will only further focus attention on technical access controls,” wrote the authors. “To the extent that any U.S. court finds an affirmative ‘fair use’ defense for AI model builders, this weakening of remedies on use will inevitably create an even stronger demand to enforce controls on access.”
The study received funding from NSF grant SaTC-2241303 and support from the Office of Naval Research project #N00014-24-1-2669.
