Horrifying stories like this should be in our minds every time we think about AI.

Pair this with Emanuel Maiberg’s article I linked to earlier, and there’s a lot to think about.

I sometimes wonder if base Grok is less wild than integrated-with-Twitter Grok, but this is at least one way in which that’s not true.

Bookmarking all these articles on Grok for rage fuel.

What a disaster.

Excellent writing on what’s going on over at Twitter.

Who could possibly have anticipated this?

Not impressed with Substack, and Masnick does a good job of explaining why.

Yesterday, I wrote my thoughts on how Grok’s “Nazi meltdown” helps illustrate some of my concerns about AI and epistemology.

This coverage of Grok’s latest demo only reinforces that—Musk’s tinkering with the LLM to get the results he wants is at odds with his states naïve epistemology that an LLM can be “maximally truth-seeking,” as though there is a self-evident truth that an LLM can deliver in a straightforward way (that is, without all that mucking about behind the scenes).

Disgusting and deliberate.

It was a real choice to start this dystopian comic the same day I got to the chapter in Karen Hao’s Empire of AI about Silicon Valley outsourcing the worst jobs to the Global South in the name of providing economic opportunity.

Interesting point from Masnick.

Honestly, I get the original call, but I’m glad Bluesky backtracked. Content moderation is hard.

This was one of the dumbest parts of all Meta’s announcements and now it’s… basically nothing, too?

Some more good writing on a bad situation.

I especially appreciate this article in the wake of Meta’s recent announcements. There are cases in which content moderation is inconsistent or overreaching, and there are cases in which less moderation fixes the problem. However, it’s the arbitrary distinctions and self-serving nature of the changes that make the “free expression” argument so flimsy.

Moderating employees but not users seems telling to me.

Personally, I’ve come around to the idea that fewer restrictions, “Community Notes”-style responses, and lots of personal control over what one sees could work better as content moderation. It’s interesting to see Bluesky take some of these approaches, for example!

Here, though, this seems like it’s bending the knee to the Trump administration, and I can’t imagine any of this being done in good faith. What a disappointment Meta continues to be.

See, this is the kind of content moderation we ought to worry about (and why “keep the kids safe” narratives can go horribly wrong).

Interesting stuff here. I think most complaints about OpenAI “censorship” are hogwash, but it’s still fascinating—and worrying—to see how much control the company exercises over its product.

Lots of interesting stuff in here, including the difficulty of content moderation, and yet another way that generative AI is screwing everything up.

Worth reading (and bookmarking). I’ve been hesitant to make the “no, actually, Big Tech is biased against liberals” argument, but this seems a compelling datum for that conclusion…

I. With Siblings in a Sibling-Only Chat Separate from the One with Parents and Partners

We process the news together (I’m not the one to start the conversation but glad for the sibling who did). It’s not a deep processing, but I’m not sure we would have done this much even a few years ago. We plan to send flowers to the widow, decide who’s going to write the note, and settle up over Venmo. We coordinate flights and talk about travel logistics, especially while our parents are waylaid with a surprise surgery that is making a hard week even harder. We talk about how other family members are doing. It shows that we’re all well into adulthood now, and that sometimes we even act like it.

This is one of these stories where I’ve been waiting to get Masnick’s take on it, and he does not disappoint.

This is a helpful EFF overview that I’m posting to bookmark for later.

Bookmarking so I have enraging examples to show my students.