AI has been trained using vast amounts of online content, which is similar to how humans acquire knowledge to create content. For instance, I use AI to assist with coding and have noticed that it can generate code that utilizes non-existent functions from non-existent libraries that appear flawless but do not actually exist. Therefore, in its attempt to be creative, AI can create inaccurate code functions that have never existed before, effectively becoming a bullshit generator. My point is that it is not merely copying someone else's work, but rather using all the information it has been fed to create something new that appears to fit the requirements. I would argue that AI using public content is not that much different from how the same content influences the creation of human content. It will be interesting to see how the laws evolve around what content the AI can use to train its future models with but my guess is ultimately anything that publicly accessible online will be fair play for AI's to consume.