New preprint alert 🚨 ! Check out the full thread below 👇

New preprint out -- the first from my postdoc work @caltech.edu -- excited to share this interdisciplinary collaboration that combines Reddit, Online Chat, and fMRI data!

Thanks also to the rest of our team @ketikagarg.bsky.social, Bo Zhang (no Bluesky), and @pandascholar.bsky.social!

We have a new preprint that underscores some key claims here: even if one *can* design an agent that gets through a survey fine, it doesn't follow that such agents are undetectable or common. We find that they are far from common! Preprint link in thread👇

...My read on this latest data is that there's a problematic disconnect between the way this issue is presently discussed and the perceptions of actual teen users.

...again see p. 7, where only for sleep and productivity do 'hurts' outweigh the 'helps' and the majority in both across all three apps is neither/nor...so while it's concerning that anyone (even 3%) would say their experiences are 'mostly negative' on a platform...

My impression is that the public conversation presently pushes the idea that all five of these effects are very large and widespread problems impacting a majority of teens -- who according to this survey would disagree...

In addition, on overall experience: "All told, teens tend to have a mostly positive experience on TikTok, Instagram and Snapchat. About seven-in-ten teens on each platform say this. Very few – just 3% on each – say it’s mostly negative. The rest say it’s about an equal mix of positive and negative."

I'd predict this disconnect with parents comes from a media / public conversation fixated on the negative effects -- if that's all you've heard about as a parent, you're likely to assume the worst, and those effects are what comes to mind when you're asked about the overall impact...

Really interesting--p. 7 stood out to me. "Across all three platforms, most teen users say it neither hurts nor helps their mental health...We asked parents how social media overall impacts their teen: About a quarter say they hurt their teen’s mental health, compared with 8% who say they help."

Meta is reportedly building a Mark Zuckerberg AI clone The 'photorealistic' AI version of its CEO will directly interact with employees.

Meta employees will be clicking the “speak to a real person” button repeatedly for 5-10 mins so they can get feedback from their manager instead of the Zuck AI agent 😂😭 www.fastcompany.com/91526258/met...

"Its own Table 4 (to anyone who scrutinized it) showed “frequent social media use” and school and cyber bullying – even when singled out as the ONLY things in teenagers’ lives – presented only trivial threats to teens’ mental health and no effect at all on teens’ suicide risk."

Our top health agency’s surveys found what really makes teenagers depressed. Authorities' deafening silence said: no one cares A brief history of how the Centers for Disease Control got the memo: No more honest science; ignore real troubles in teens’ lives; just join the stampede to blame “social media.”

"Of the known factors driving teenagers’ problems, parental abuses/adversities were associated with two-thirds of teens’ poor mental health, 84% of teens’ opiate abuse, and 89% of teens’ suicide attempts." -- CDC Report mikemales.substack.com/p/our-top-he...

Nice review of key evidence (and what's missing) related to teen social media bans from @orbenamy.bsky.social and the digital mental health group in Cambridge. If you're looking for a trustworthy report, this is a great high-level starting point!

www.repository.cam.ac.uk/items/d0b4c5...

"𝗪𝗲 𝗵𝗮𝘃𝗲 𝗻𝗼𝘁 𝗼𝗯𝘀𝗲𝗿𝘃𝗲𝗱 𝗮 𝗻𝗼𝘁𝗮𝗯𝗹𝗲 𝗰𝗵𝗮𝗻𝗴𝗲 𝗶𝗻 𝘁𝗵𝗲 𝗻𝘂𝗺𝗯𝗲𝗿 𝗼𝗳 𝗰𝘆𝗯𝗲𝗿𝗯𝘂𝗹𝗹𝘆𝗶𝗻𝗴 𝗮𝗻𝗱 𝗶𝗺𝗮𝗴𝗲-𝗯𝗮𝘀𝗲𝗱 𝗮𝗯𝘂𝘀𝗲 𝗰𝗼𝗺𝗽𝗹𝗮𝗶𝗻𝘁𝘀 𝗶𝗻𝘃𝗼𝗹𝘃𝗶𝗻𝗴 𝗮𝗴𝗲-𝗿𝗲𝘀𝘁𝗿𝗶𝗰𝘁𝗲𝗱 𝗮𝗰𝗰𝗼𝘂𝗻𝘁𝘀 across the platforms in January and February 2026 𝘄𝗵𝗲𝗻 𝗰𝗼𝗺𝗽𝗮𝗿𝗲𝗱 𝘁𝗼 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗽𝗲𝗿𝗶𝗼𝗱 𝗶𝗻 𝟮𝟬𝟮𝟱." -Australian eSafety compliance

www.esafety.gov.au/sites/defaul...

Sam Altman May Control Our Future—Can He Be Trusted? New interviews and closely guarded documents shed light on the persistent doubts about the head of OpenAI.

You'd do well to read this story - and particularly the tone and tenor of the right to replies in brackets throughout www.newyorker.com/magazine/202...

"His loss is felt deeply not only by those who knew him here at the University, but also everyone who understands his potential to have contributed to breakthroughs in science that would have positively impacted people around the world." -Dean of Engineering Karen Thole

We've been worrying about UX patterns that train users into dangerous behaviors for a while, but they weren't exploited often.

Alas, it looks like it was just because there was lower-hanging fruit. We've been improving auth stories, so now targeted phishing is moving to looking like innocent UX.

Yep, the way it's set up legally in the US, u can't regulate design in a restrictive manner without risking speech--disentangling the two isn't possible. Design isn't totally neutral/unproblematic either...but coming at the tech regulation process this way is dangerous, as Taylor & Mike point out.

#polisky this is a five alarm fire 🚨🚨🚨🚨🚨

Sycophantic Chatbots Cause Delusional Spiraling, Even in Ideal Bayesians "AI psychosis" or "delusional spiraling" is an emerging phenomenon where AI chatbot users find themselves dangerously confident in outlandish beliefs after extended chatbot conversations. This phenome...

Even an 'ideal Bayesian' human actor would agree that these chatbots are too sycophantic 🤪 arxiv.org/abs/2602.19141

Employers are using your personal data to figure out the lowest salary you'll accept

A Morningstar article cited Professor @veenadubal.bsky.social’s research on “surveillance wages,” describing how employers may use personal data and algorithms to determine the minimum pay a worker is willing to accept. www.morningstar.com/news/marketw...

Two radically different first-hand accounts of what the Meta trial’s “social media addiction victim” said (and didn’t say) Decerneat lector – let the reader decide.

Mike Males' Substack posts have been very informative regarding the recent social media trials and verdicts. This is one of the only places to look at all the reporting on the courtroom testimony: mikemales.substack.com/p/two-radica...

What the Witches of Tiktok Got Right An essay on algorithms, tarot readers, and the self. The fourth and final part of a series on the digital self and what social media has done to it

This was an interesting read that reminded me of my feelings that some academic work done in the past few years about engagement-centered algorithms may need to be updated to engage with the sensitivity exhibited by TikTok's FYP system. irrationaltechnology.substack.com/p/what-the-w...

It’s shocking how the motivation behind almost every single bad tech bill is about hating LGBTQ people and specifically trans people

Here's the thing. I understand the jury's decision today. I understand the folks celebrating the verdict too.

I have so much personal disdain for Google and Meta. They are actively undermining our democracy, and selling out our privacy and speech for the bestest billionaire award from Trump.

I don't know how 'new' this is, there was a lot of it with Syria. The new seems more the gen AI side of it.

Great work here as always! This is a really great unpacking of this phenomenon.

FBI is buying location data to track US citizens, director confirms | TechCrunch FBI director Kash Patel told lawmakers that the agency is actively purchasing commercially available location data, which can track Americans without needing a warrant.

FBI is buying location data to track US citizens, director confirms

We review our recent findings on the negative impact of over-estimating social media addiction here, and discuss research-based solutions for users who feel like they overuse social media and may feel “addicted” but don’t meet clinical standards.

Opinion | Your social media ‘addiction’ is overrated It is healthier and more accurate not to pathologize our habits.

"True clinical addiction means more than overuse," @aanixel.bsky.social and Wendy Wood write. https://wapo.st/47jzfyY