AI companies are suddenly scrambling to watermark their work. Anthropic said this week it will start marking text created by its AI, including Claude, so it can be identified as AI-generated. The move complies with a new European Union law that kicked in August 2, according to TechCrunch. Every new model going forward will have this built in automatically, and the mark is supposed to stick around even after you copy and paste the text somewhere else.

It’s still unclear exactly how much you’d need to edit that text before the mark disappears. TechCrunch asked Anthropic point-blank and didn’t get a straight answer. Anthropic says it’ll eventually add this to older models too, but hasn’t said when.

Anthropic isn’t the only one doing this. AI music app Suno said last week it’ll start tagging songs made on its platform after getting hit with lawsuits. Newsletter platform Substack teamed up with a detection company last month to flag AI-written content, after its CEO complained about people passing off AI writing as their own, something he calls “Claudefishing.” Big names like Google, Meta, Microsoft and OpenAI have all agreed to do the same.

AI companies are suddenly scrambling to watermark their work. Anthropic said this week it will start marking text created by its AI, including Claude, so it can be identified as AI-generated. The move complies with a new European Union law that kicked in August 2, according to TechCrunch. Every new model going forward will have this built in automatically, and the mark is supposed to stick around even after you copy and paste the text somewhere else.

It’s still unclear exactly how much you’d need to edit that text before the mark disappears. TechCrunch asked Anthropic point-blank and didn’t get a straight answer. Anthropic says it’ll eventually add this to older models too, but hasn’t said when.

Anthropic isn’t the only one doing this. AI music app Suno said last week it’ll start tagging songs made on its platform after getting hit with lawsuits. Newsletter platform Substack teamed up with a detection company last month to flag AI-written content, after its CEO complained about people passing off AI writing as their own, something he calls “Claudefishing.” Big names like Google, Meta, Microsoft and OpenAI have all agreed to do the same.