Out with the screens, in with the books, or maybe we land somewhere in between.

For American Public Media, this is Marketplace Tech. I'm Stephanie Hughes.

Time for Marketplace TechBites, your regular roundup of a few tech stories.

This week, Spotify is letting its users buy physical books through its app.

Plus, Amazon acquires a satellite service provider Global Star.

But first, state lawmakers in Illinois are considering a bill that says developers of large AI models

can't be held liable for critical harms caused by those models.

That's as long as the developer doesn't intentionally or recklessly cause the harm

and has published a safety protocol on its website.

A representative from OpenAI testified in favor of the bill.

Meanwhile, Wired reported this week that Anthropic is pushing for either major changes to the legislation or for it to be killed completely.

I asked Maria Curie, tech policy reporter at Axios, why we're seeing this split between companies.

You know, I think that at the federal level, it's very unlikely that any

legislation is going to move either on the side of a pro-innovation deregulatory move that preempts

these types of state level measures or on the flip side, a transparency, safety first type measure,

which is, you know, where anthropic fits more toward. And so what we're

seeing instead is states taking the lead and these companies kind of going in and either playing

whackamol or trying to support various legislation at the state level in the absence of action

here in Washington, D.C. The bill would let companies off the hook for extreme harms caused by their AI models.

This is essentially treating harm by artificial intelligence differently from harm caused by humans.

Why do you think we're seeing this approach?

You know, it's not totally new if you think about the way that we treat social media companies

...