Daphne Keller’s Post

I can't express how tired I feel after reading this absurd 3rd Circuit ruling. It denies TikTok 230 immunity for a claim that is (very thinly) framed as liability based on algorithmic promotion of user content, instead of liability based on user content. https://lnkd.in/gBTh4dGx This issue was fully briefed to the Supreme Court, with approximately infinity amicus briefs covering every possible angle, a little over a year ago. The Court decided not to decide. https://lnkd.in/gihJ5Rgm Now the 3rd Circuit is engaging in the absurd pretense that the Court actually decided this issue in Moody v. NetChoice, because it said algorithmic ranking that advances the platform's content moderation goals is the platform's 1st Am protected speech. (Check out fn 13 in the TikTok case, sigh.) There is NO CONFLICT between moderation and ranking being (1) the platform's speech and also (2) immunized by 230. The whole point of 230 was to encourage and immunize moderation. As the law's drafters pointed out, that includes algorithmic ranking. https://lnkd.in/g-vMtTHm If Section 230 did not apply to platforms' constitutionally protected exercise of editorial freedoms, then it would apply to basically nothing. It would be a nullity. Seriously how can we still be having this conversation. How can the 3rd Circuit believe its own nonsense on this. How dare they create a circuit split and force an exhausted nation to go through this AGAIN.

I spent an hour on Friday talking with Mike Masnick about the week's news. That included two different 9th Circuit rulings, important developments in Brazil, and the things I learned about platform regulation from ChatGPT. https://lnkd.in/gNnqgR-t

This was such a fun podcast conversation, and great set of topics! It’s exclusively about stuff from the past week, which I thought would be limiting. But there is SO much going on.

I had some remarkable conversations with ChatGPT last week, and I've written a blog post about them. The post is about AI's remarkable capacity for saying the quiet part out loud. It spelled out in no uncertain terms that euphemistic rules like the "design" mandates in the Kids Online Safety Act (KOSA) require platforms to suppress legal speech. https://lnkd.in/gVXWa55n I think that is clearly true of KOSA, as it is of California's Age Appropriate Design Code. ChatGPT also said that the EU's Digital Services Act requires the same thing -- in that case, I am holding out hope that it's wrong. This isn't an example of AI being quirky or startlingly wrong or telling us to eat glue. This is AI reflecting what its training materials say, including what politicians, news reports, and secondary sources say about these laws on the Internet. It is articulating a perfectly plausible interpretation. *This is the interpretation we should expect compliance-minded platforms to apply.* If removal of lawful speech is not the goal, then we shouldn't pass laws like these. https://lnkd.in/gVXWa55n

NetChoice v Bonta, the California Age Appropriate Design Code case, came down from the Ninth Circuit today. This was IMO a pretty easy on the substance, and pretty hard in needing to apply the Supreme Court's NetChoice v. Moody facial/as applied logic to a big sprawling law. The 9th Circuit did a very good job on both. https://lnkd.in/gqCeVEBV The easy part was about the law's requirement to assess risks that were *defined by the kinds of content* children might see, and then mitigate those risks. Obviously that is a speech regulation. The panel gave the AG lawyer a hard time for trying to pretend otherwise. The fact that this mandate was called a "Data Protection Impact Assessment" and talked about both things at once doesn't change this bottom line. As the court notes (later, in tailoring analysis), lawmakers coulda just made a law about data and not pinned it to speech. And as the court also notes, the fact that this content-based law is unconstitutional tells us zip, zero, nada, zilch, bagel, donut about whether ACTUAL privacy laws are OK. You know, the ones that don't require straight-up speech restrictions. This whole part of the law -- conducting the DPIA, writing it up, mitigating risks of kids seeing proscribed content -- is a content-based restriction. This is not commercial speech, the law gets strict scrutiny, and it fails. As do a handful of related administrative parts. One reason it fails is the reason Jack Balkin and I wrote about in our Moody brief for Francis Fukuyama: BECAUSE EMPOWERING USERS WOULD BE A LESS RESTRICTIVE MEANS. Also maybe enforcing existing laws against bad guys! https://lnkd.in/g9kqHmex hen the court says it cannot facially invalidate some remaining provisions, and those need much closer analysis. Several of these are what I would consider normal privacy laws. No real speech issue. But a few are Big Deals that the 9th was surely happy to dodge. In that "unresolved and actually a big deal" list are (a) the age assurance requirement, which uses carefully wiggly language but is at least (unlike in KOSA) acknowledged to exist. (b) the requirements to publish a whole bunch of info in language understandable by kids of a (presumably often very wide) range of ages, and to enforce platform speech rules in... whatever way the AG thinks they should be enforced, I guess? (c) the "dark patterns" rule, which in the statute looks like a regular old circa 2018 privacy-based use of the term, but which "design expert" amici told the court includes things like infinite scroll? Anyhow the 9th is like we don't know what this even is, we are tired, send it back. (Will continue in comments, just a bit more!)

In case anyone missed it, there's a new Justice Thomas anti-230 screed, this time with Gorsuch joining. Eric Goldman breaks it down here. https://lnkd.in/g3D3AHwE

I clicked through to answer “skepticism.” But this turned out to be a multiple choice question, and that wasn’t one of the answers.