Lawyers are paying close attention to the proposed regulation of generative artificial intelligence in Europe, the US and China as their clients contend with the unclear risks of violating intellectual property rights.

“If I need to comply with both US and EU law, and one of them lets me do it, and the other one doesn’t, I still can’t do it,” says Mark Lemley, a Stanford Law School professor and counsel at Lex Lumina, which is defending London-based Stability AI and other generative AI companies against copyright infringement claims filed by picture agency Getty Images.

A “bigger danger” looms for users of GAI if the output appears “too similar to something . . . on [the] input side,” Lemley says. But that only happens with existing GAI systems if the user asks for the output to be similar to a specific input, he notes.