Casey Newton
LightNews LightNews
banner
caseynewton.mastodon.social.ap.brid.gy
Casey Newton
@caseynewton.mastodon.social.ap.brid.gy
140 followers 0 following 160 posts
Email salesman at Platformer and podcast co-host at Hard Fork. casey@platformer.news

[bridged from https://mastodon.social/@caseynewton on the fediverse by https://fed.brid.gy/ ]
Posts Replies Media Videos
caseynewton.mastodon.social.ap.brid.gy
A novel new study finds you can reduce polarization on X simply with a simple change to the algorithm. I spoke with the researchers: https://www.platformer.news/stanford-polarization-study-x-algorithm/
The results: reducing exposure to divisive content improved feelings toward the other party by about 2 points on a 100-point scale. On the other hand, boosting that content cooled feelings by about 2.5 points. The researchers say the shift is equivalent to about three years of rising partisan animosity in America, created (or reversed) in just a week. Remarkably, researchers found this result without adding or removing hyper-partisan material — it simply changed the order in which the posts appear. (For people who saw less polarizing content, for example, it was still in their feeds — just way lower, and thus less likely to be seen.) “I think the general intuition is that there's a little bit of a failure of imagination going on with a lot of the platforms,” Michael S. Bernstein, a researcher at Stanford and co-author of the paper, told me in an interview. “We think they were sort of painted into a box where they felt like they had to optimize for certain engagement criteria, like that's the only thing they could do. It just felt like that didn't necessarily have to be the case anymore."
December 2, 2025 at 12:59 AM
caseynewton.mastodon.social.ap.brid.gy
Record labels had the chance to stand up for copyright against AI scrapers — but now they're folding https://www.platformer.news/suno-warner-settlement-copyright-udio-klay/
The move toward settlements and licensing agreements, distasteful as it no doubt seems to many artists, represents a safer path. And so long as artists can opt out of their work being included, the most obvious direct harm to individuals can be avoided. Still, you can probably imagine what lesson the tech industry will take from this development. Just as Google learned with books, or YouTube did with Viacom, it rarely pays to ask for permission — and forgiveness often comes more cheaply than you would guess.
November 26, 2025 at 12:59 AM
caseynewton.mastodon.social.ap.brid.gy
I wrote about our interview with Roblox's CEO and how platforms got too big to feel shame https://www.platformer.news/roblox-ceo-interview-backlash-analysis/
Fast forward to today and the bargain no longer holds. Platforms do whatever the president of the United States tells them to do, and very little else. Shame, that once-great regulator of social norms and executive behavior, has all but disappeared from public life. In its place is denial, defiance, and the noxious vice signaling of the investor class. I'm still reckoning with what it means to do journalism in a world where the truth can barely hold anyone's attention — much less hold a platform accountable, in any real sense of that word. I'm rethinking how to cover tech policy at a time when it is being made by whim. I'm noticing the degree to which platforms wish to be judged only by their stated intentions, and almost never on the outcomes of anyone who uses them.
November 25, 2025 at 2:52 AM
caseynewton.mastodon.social.ap.brid.gy
It is strange but great to see so many Republicans turning against Trump over AI safety concerns https://www.platformer.news/trump-ai-moratorium-republican-backlash/
And why shouldn’t they? AI is not particularly popular, and Americans are increasingly concerned about its potential effects on their jobs, their children, and energy prices. Silicon Valley won the president’s favor by paying him millions of dollars and creating crypto products that have enriched him and his family. It has not done the same for Republicans more broadly. And before too long, being seen as a “pro-AI candidate” may prove a more toxic brand than “pro-crypto candidate” ever was. Trump’s effort to smother AI regulations despite overwhelming opposition from the public and Congress offers yet another reminder of his willingness to auction his policy positions to the highest bidder. But the backlash illustrates that most American politicians still show real concern about what their constituents think. And most Americans who have followed the issue believe that Congress got this one right the first time.
November 21, 2025 at 1:23 AM
caseynewton.mastodon.social.ap.brid.gy
In the end, all it took down to bring down the government's case against Meta was a simple experiment https://www.platformer.news/ftc-loses-meta-case-analysis/
John List, a professor at the University of Chicago, recruited 6,000 participants for a study. With permission, he installed tracking software on their phones. And then he made them an offer: for every hour that the participant did not use Facebook or Instagram, he would pay them $4. (Participants only had to give up one app; the Facebook holdout group was allowed to use Instagram and vice versa.) Most participants were happy to stop using Meta products for a few dollars; their usage fell by about two-thirds. More important, though, is what they did on their phones without Facebook or Instagram. And it wasn’t using MeWe.
November 19, 2025 at 1:53 AM
caseynewton.mastodon.social.ap.brid.gy
Are there any *good* reasons to consider engagement-maximizing algorithms protected speech? I wrote about the California law that has platforms shaking https://www.platformer.news/california-algorithm-speech-lawsuit-meta-google-tiktok/
In my layperson's understanding of the law, calling algorithmic predictions about which posts a person might like or leave a comment on "speech" feels like a corruption of the term. Particularly because no person at any of these companies can tell you with any certainty why any individual post appeared in your feed. They can tell you what signals they took into account, and which of your past behaviors likely influenced what you're seeing. But they also constantly show people posts that they later express regret over — porn, scams, violence and so on — and profess ignorance when asked why their systems showed those posts to users. It's a funny kind of speech — one that can be traced to no individual person, and is in a language that no one fully understands.
November 18, 2025 at 1:41 AM
caseynewton.mastodon.social.ap.brid.gy
I wrote about Europe maybe retreating from strong privacy protections in the name of AI competitiveness and why it may be making a mistake https://www.platformer.news/eu-ai-act-changes-proposal/
It would be one thing if onerous AI regulations were preventing EU citizens from getting lifesaving medical care or high-performing AI coworkers. For the moment, though, Europe is getting just about everything Americans are — a few weeks to a few months later. (And for everything else, there are VPNs.)
November 11, 2025 at 1:33 AM
caseynewton.mastodon.social.ap.brid.gy
Wrote about Perplexity's comical flouting of every other website's rules and insistence that it is being "bullied" https://www.platformer.news/amazon-perplexity-comet-agents-future/
Amazon will be fine regardless of how this particular conflict plays out. But I’m curious how much longer Perplexity will be able to keep up its act. The company loves to present itself as an underdog. But you can only slap so many people in the face before they start asking if the real bully here might be the one doing all the slapping.
November 5, 2025 at 2:12 AM
caseynewton.mastodon.social.ap.brid.gy
I wrote about Bluesky a year after the big post-election surge in users, and whether you can build a happier version of Twitter while keeping its same basic shape https://www.platformer.news/bluesky-2025-content-moderation-fediverse/
t's clear that you can build a nicer online environment than X has; in many ways Bluesky already did. What's less clear is that you can build a Twitter clone that mostly makes people feel good. For as vital and hilarious as Twitter often was, it also accelerated the polarization of our politics and often left users feeling worse than they did before they opened it. Bluesky's ingenuity in reimagining feeds and moderation tools has been a boon to social networks, which have happily adopted some of its best ideas. (You can now find "starter packs" on both Threads and Mastodon.) Ultimately, though, it has the same shape and fundamental dynamics as a place that even its most active users called "the Hellsite."
November 4, 2025 at 1:53 AM
caseynewton.mastodon.social.ap.brid.gy
Character.AI is the first major company to acknowledge that companions aren't safe for young people. Will others follow? https://www.platformer.news/character-ai-teen-ban/
Maybe Character.AI banned teens from using its companions simply because executives believed it is the right thing to do. Or maybe they became convinced that the company’s mounting legal liability had forced their hand. Whatever the case, in a benighted year for trust and safety, it’s at least somewhat heartening to see a company shrink itself in the name of protecting its users. At a time when AI is presented as an inevitability, Character.AI’s retreat offers a timely reminder that human beings are still in control. You don’t have to build it; you don’t have to release it; and you can still pull the plug. Here’s hoping other companies heed its example.
October 31, 2025 at 12:05 AM
caseynewton.mastodon.social.ap.brid.gy
Wrote about OpenAI's restructuring and what it means for how AI is governed https://www.platformer.news/openai-restructuring-microsoft-altman/
The optimistic view of these changes is that pushback from activists and regulators resulted in a meaningful oversight mechanism being preserved within the nonprofit that runs OpenAI. The pessimistic view is that while the mechanism exists on paper, recent history suggests that this particular board takes a fairly hands-off approach to corporate governance. In one sense, yes, the nonprofit still runs the for-profit. But the board is stocked with entrepreneurs, including Altman and at least two people (Taylor and Adam D’Angelo) who run businesses that are customers of OpenAI. Is the nonprofit truly running the for-profit — or is it the other way around?
October 29, 2025 at 12:07 AM
caseynewton.mastodon.social.ap.brid.gy
I talked to OpenAI about its surprising new research on the prevalence of mental health concerns in ChatGPT usage and whether this is a problem we can model-tweak our way out of https://www.platformer.news/openai-mental-health-research-chatgpt-suicide-delusions/
Tweaking a model’s output may well reduce self-harm and delusions in a chatbot’s users. But whatever changes OpenAI or other companies make, they will also simultaneously be working to build products that users increasingly feel like they can’t live without. Even as it works to reduce emotional dependence on ChatGPT for some subset of distressed users, OpenAI will be working to increase users’ overall dependence on its products for work, entertainment, web browsing and more. And it will do so in part by reliably telling us what we want to hear. It would be great if AI companies continue to see emotional dependence on their products as a problem to be solved. But it seems likely that, at least to some of them, it will eventually be evidence that the product is working as intended.
October 28, 2025 at 12:10 AM
caseynewton.mastodon.social.ap.brid.gy
OpenAI's flailing over content policy for Sora is just the latest example of the company rushing out a product at the expense of its own reputation. As AI tools get more powerful, that's reason for concern https://www.platformer.news/openai-sora-copyright-policy-king-cranston/
It’s just as clear that this approach has long stopped serving the company well. From the Scarlett Johansson voice mode controversy to GPT-4o sycophancy to the more recent backlash against GPT-5, OpenAI’s rushed release schedule now routinely puts the company on the defensive. But with Altman now openly acknowledging the connection between revenue pressures and product releases, there is every reason to believe that policy concerns will continue to be eclipsed by the urgency of shipping.
October 21, 2025 at 12:17 AM
caseynewton.mastodon.social.ap.brid.gy
I wrote about the AI slop backlash and how platforms are starting to respond to it https://www.platformer.news/thursday-newsletter-5/
The popularity of surreal AI genres such as "slicing through glass fruit," Italian brainrot, and shrimp Jesus helps to explain much of the disdain for them. If slop weren't occasionally compelling, we might feel like we could safely ignore it. But to some relatively large audience, they are, the 3 million views on this AI clip of a knife slicing through glass pancakes and waffles attest. And so if you're the sort of person who prefers your internet man-made — or at least in healthy balance with AI models — the rise of slop understandably feels like a threat. Increasingly, a coalition of average people and copyright holders are pushing the platforms themselves to take action — and they're beginning to see some success.
October 17, 2025 at 12:58 AM
caseynewton.mastodon.social.ap.brid.gy
California's move to regulate AI companions really could not have come soon enough https://www.platformer.news/california-ai-companion-law-newsom-openai/
For the moment, chatbot companions are mostly seen as a sideshow. That misses just how quickly they are growing in popularity. An analysis by ChinaTalk this month found that AI chatbots designed for sex and romance have about 29 million monthly users globally. (That's about 6 million more people than use Bluesky.) Meanwhile, a recent survey by the Center for Democracy and Technology found that one in five high school students say that they or someone they know has been in a romantic relationship with a chatbot. Those data points pair uneasily with statements from the CEOs of AI labs promising to make their products simultaneously more personable, more sexual, and more powerful. Some segment of the population seems already to be coming to the conclusion that AI companions are more engaging, agreeable, and easier to talk to than people. And the response of many chatbot makers so far has been to take those qualities and supercharge them.
October 15, 2025 at 12:28 AM
caseynewton.mastodon.social.ap.brid.gy
AI leaderboards are declining in influence. So why did Meta-backed Scale AI just launch a new one? https://www.platformer.news/ai-leaderboards-decline-lmarena-scale/
But over the past year, crowd-voted leaderboards have become a much less reliable way to judge the quality of chatbots. Most LLMs are already more than good enough for casual chat, the leaderboards have often been gamed, and top leaderboards now disagree on which model is best. Inside labs and among power users, benchmarking has shifted to task-specific evaluations and other niche testing — tools designed to predict real-world performance rather than put the question up to a vote.
October 10, 2025 at 12:14 AM
caseynewton.mastodon.social.ap.brid.gy
Wrote about fresh signs of an AI bubble (and why the industry says the worries are mostly overblown) https://www.platformer.news/ai-bubble-2025/
We have entered the "weird financial tricks" phase of the bubble. A reliable indicator of bubbles is that companies concoct bizarre financial instruments and accounting measures to paper over their problems. In Fortune, Allie Garfinkel reports that venture capitalists are irked at some of the creative accounting going on in their portfolios. Some startups are reporting one-time sales as "recurring revenue." Others technically show "recurring revenue" for what are essentially trials. And that's on the tame side. For crazy, you can look to Robinhood's suggestion that it might offer "tokenized" equity in OpenAI. By "tokenized," Business Insider reports, Robinhood means "blockchain-enabled representations of securities like stocks." In reality, they have no connection to OpenAI equity whatsoever. But that doesn't mean that consumers shouldn't be able to gamble on the mirage!
October 8, 2025 at 12:30 AM
caseynewton.mastodon.social.ap.brid.gy
In this edition we're also launching the Following feed, part of our new approach to curating links. Each day we're giving you three to five stories we're following, why we're following them, and what people are saying about them. Today we wrote about reactions to Sora
Zelda Williams, daughter of the late comedian Robin Williams, begged people in an Instagram story to stop sending her Sora videos of her dad. "You're not making art, you're making disgusting, over-processed hotdogs out of the lives of human beings, out of the history of art and music, and then shoving them down someone else's throat hoping they'll give you a little thumbs up and like it," she wrote. "Gross." Meanwhile, what do average users want? Fewer content restrictions. We found App Store reviews reading “It’s so censored it’s not even fun,” and “the “safeguards in place prevent you from making anything worth watching,” among other complaints.
October 6, 2025 at 10:48 PM
caseynewton.mastodon.social.ap.brid.gy
I wrote about OpenAI's ambitious platform play and the Cambridge Analytica flashbacks it inspired. Plus, Sam Altman takes my question about enshittification risk https://www.platformer.news/openai-dev-day-2025-platform-chatgpt/
At launch, OpenAI is promising a more rigorous approach to data privacy. OpenAI will share only what it needs to with developers, executives said. (They essentially hand-waved through the details, though, so the actual mechanics will bear scrutiny.) Unlike Facebook, though, OpenAI has no friend graph to worry about — whatever might go wrong between you, ChatGPT, and a developer, it will likely not involve giving away the contact information of all of your friends. At the same time, the AI graph may prove even riskier. ChatGPT stores many users’ most private conversations. Leaky data permissions, either intentional or accidental, could prove disastrous for users and the company. It only took one real privacy disaster to end Facebook’s platform ambitions; I can’t imagine it would take much more to end OpenAI’s.
October 6, 2025 at 10:45 PM
caseynewton.mastodon.social.ap.brid.gy
Here's what everyone is saying about Sora. Of note: several OpenAI employees have posted about their concerns with the product https://www.platformer.news/sora-2-reactions-openai/
It makes OpenAI employees nervous. One of the more surprising reactions to the launch came from OpenAI's own employees, several of whom arched an eyebrow about the company's latest product. "AI-based feeds are scary," posted John Hallman, who works on pretraining at the company. "I won't deny that I felt some concern when I first learned we were releasing Sora 2. That said, I think the team did the absolute best job they possible could in designing a positive experience." (Here's a funny response to that one.) "I share a similar mix of worry and excitement," responded Boaz Barak, a member of OpenAI's technical staff. "Sora 2 is technically amazing but it’s premature to congratulate ourselves on avoiding the pitfalls of other social media apps and deepfakes."
October 3, 2025 at 12:04 AM
caseynewton.mastodon.social.ap.brid.gy
Tried out the new slop app https://www.platformer.news/sora-2-hands-on-openai-social-network/
And what if it is the answer? What if perfectly personalized entertainment takes over our screens and our socializing? The arrival of Sora has made those questions feel suddenly more urgent. This is the first social network that — on launch day — said it will disable infinite scroll for teens, and nudge adults to put their phones down when they've been browsing for too long. It's a responsible step for OpenAI to take. But it might better be viewed as a warning.
October 1, 2025 at 12:29 AM
caseynewton.mastodon.social.ap.brid.gy
Talked to OpenAI about its new parental controls and the effort to make ChatGPT safer for kids — by pushing more responsibility onto parents https://www.platformer.news/chatgpt-parental-controls-child-safety/
You can appreciate the new choices that OpenAI is giving parents here while also noting that parental controls also push at least some of the responsibility for ChatGPT safety onto those same parents — while also requiring those same parents to create ChatGPT accounts of their own. In that, OpenAI joins Meta, Snap, TikTok, and other apps popular with teenagers that all but require parents to take on a part-time job learning and managing the settings of apps that change continuously and introduce new risks as they do. At the same time, this is a feature that could absolutely save lives. It’s also a feature that no other chatbot maker has yet introduced, despite high-profile cases of harm on their platforms.
September 30, 2025 at 12:10 AM
caseynewton.mastodon.social.ap.brid.gy
ChatGPT Pulse is a step toward becoming a proactive assistant that works for you in the background. It's also one more feed to scroll. Sound familiar? https://www.platformer.news/chatgpt-pulse-proactive-ai/
OpenAI says it wants ChatGPT to be helpful rather than addictive, and that’s the way I experience it today. But a feature like Pulse, which looks and feels like a social feed, could create other incentives. (Particularly given that the company is reportedly now also looking for an ads chief, as Alex Heath reported in his new newsletter Sources this week.) Notably, many OpenAI executives and product leaders came from Meta, including new CEO of applications Fidji Simo. Over time, Facebook’s growth imperatives led it to greatly expand the reasons why it would send you a push notification, until eventually they had no real connection to the user at all. (Today Facebook notified me that someone had listed a 1987 Dodge Ram on Marketplace, a feature that I have never once used.)
September 25, 2025 at 10:42 PM
caseynewton.mastodon.social.ap.brid.gy
Today is Platformer's fifth birthday! I wrote about the state of the newsletter economy, how the business is going, and what we're planning for year six https://www.platformer.news/platformer-year-five-lessons/
Three, we’re going to create an audio feed. For years now, readers have asked me to narrate each edition so you can listen to it as a podcast. (As pioneered by Ben Thompson’s great Stratechery.) I’ve resisted until now because I worried the added work would burn me out. (Candidly, I still do.) But the success of Hard Fork — one of the 100 best podcasts of all time, by the way! — has introduced our work to a lot of people who would rather listen to me than to read. I’m still in the early stages of planning this, and I expect to encounter many technical challenges along the way. But my hope is that a year from now you’ll get regular audio drops from Platformer that make your subscription feel valuable.
September 24, 2025 at 12:07 AM
caseynewton.mastodon.social.ap.brid.gy
I wrote about how social media enabled Charlie Kirk while making our politics ever worse. https://www.platformer.news/thursday-newsletter-4/
X remains the platform of choice for most politicos, and continues to play an outsize role in shaping elite perception of our politics. The dark picture of America that takes root there spreads to TikTok, Instagram, and cable news, continuously warping Americans' perceptions of one another. The vast majority of Americans continue to reject political violence. Civil war is for idiots and losers. But in the dark mirror of social media, we are only ever on the brink of all-out conflict. And the longer we stare at it, the more we risk that fiction turning into reality.
September 19, 2025 at 1:29 AM