copyright management information
The ownership and credit details attached to a copyrighted file, which US law makes it a separate offence to strip out.
Copyright management information, usually shortened to CMI, is the label attached to a copyrighted work: the title, the author, the copyright notice, the licence terms, and any identifying number. It can sit in a file’s metadata or be printed on the page.
Under section 1202 of the US Digital Millennium Copyright Act, deliberately removing or altering that information to conceal infringement is its own violation, separate from copying the work. That matters in AI cases because preparing text for training usually means cleaning it, and cleaning strips exactly this kind of label. Publishers suing AI companies now routinely claim damages twice over, once for the copying and once per stripped label, which is how the totals reach the billions.