You can enable subtitles (captions) in the video player
Content creators are striking back against tech companies brazenly harvesting their material to train AI models. Over the past year numerous news and content providers, including Getty Images, Universal Music, George RR Martin, and the New York Times have filed lawsuits against AI companies. They allege their work is being used without their permission to train the artificial intelligence that could ultimately put them out of business.
The allegations aren't baseless. A recent investigation by Proof News found that tech companies, including Apple, Nvidia, and Anthropic, have lifted material from tens of thousands of YouTube channels without permission. Another investigation last year found that more than 180,000 books had been surreptitiously used to train AI applications.
Tech giants, however, have argued that harvesting content without permission isn't illegal or breaching copyright because the material they lift is transformed and ultimately used for a different purpose than originally intended. Their supporters also contend that individuals in search of inspiration don't need permission from content creators to listen to a song or look at a painting. So why should AI robots?
They also state that given the size of these AI models, any single person's contribution to the AI output is minuscule. Some content providers have taken a more conciliatory approach, but others are taking the fight to AI. Computer scientists at the University of Chicago have recently launched two tools, called Glaze and Nightshade, which can distort or poison digital content in such a way that it disrupts AI training and scraping processes.
Meanwhile, a new US Senate bill, the Copied Act, aims to set new transparency guidelines for marking, authenticating, and detecting AI-generated content. Its goal is to protect content creators against what it calls AI-driven theft. Ultimately, tech companies' right to harvest content with impunity will likely depend on judges' interpretation of copyright law.
They may deem that AI scraping could be viewed predominantly as learning, and therefore fall within fair use. If that is the case, content creators may just need to accept what they feel is the blatantly unfair use of their material.
