Returning from winter break, this episode of the Cyberlaw Podcast covers a lot of ground. The story I think we'll hear the most about in 2024 is the remarkable exploit used to compromise several generations of Apple iPhone. The question we'll be asking is simple: How could an attack like this be introduced without Apple's knowledge and support? We don't get to this question until near the end of the episode, and I don't claim great expertise in exploit design, but it's very hard to see how such an elaborate compromise could be slipped past Apple's security team. The second question is which government created the exploit. It might be a scandal if it were done by the U.S. But it would be far more of a scandal if done by any other nation.
Jeffery Atik and I lead off the episode by covering recent AI legal developments that simply underscore the obvious: AI engines can't get patents as "inventors." What's more interesting is the possibility that they'll make a whole lot of technology "obvious" and thus unpatentable. Speaking of obvious, claiming that companies violate copyright when they train AI models on New York Times content requires a combination of arrogance and cluelessness that can only be found at, well, the New York Times.
Paul Stephan joins us to note that the National Institute of Standards and Technology (NIST) has come up with some good questions about standards for AI safety.
