Meta reportedly used copyrighted content to train its AI models and seemingly tried to cover its tracks to avoid copyright infringement-related issues, including removing copyright headers and document identifiers.
Source: Read More / Windows Central