Kevin Klyman

A screenshot of the title of the paper, "Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers."

🧵I'm thrilled to announce that I'll be going to @facct.bsky.social this June to present timely work on why current LLMs cannot safely **replace** therapists.

We find...⤵️

Our latest brief highlights the importance of independent evaluation for AI safety and accountability, addressing barriers and proposing safe harbors to protect third-party research. @klyman.bsky.social @shaynelongpre.bsky.social @sayash.bsky.social @peterhenderson.bsky.social

We need a federal data privacy law now.

Or we’ll continue down the road of a piecemeal strategy of banning popular foreign apps, while the US data broker industry continues to monetize the data of all Americans, mostly without our consent or notice. #AIBillofRights

Here's my "fact check" on Meta's announcement that it is terminating its fact checking program. The TL;DR: Pay more attention to product changes than to political pandering. THREAD...

Someone Made a Dataset of One Million Bluesky Posts for 'Machine Learning Research' A Hugging Face employee made a huge dataset of Bluesky posts, and it’s already very popular.

New from 404 Media: Bluesky may have said it won't use user data to train generative AI, but someone else just published a dataset of million Bluesky posts for "machine learning research". Already very popular dataset, your data may be scraped www.404media.co/someone-made...

Content Moderation on Bluesky The growth of Bluesky over the past couple of weeks has been spectacular. The app now has 3.5 million active daily users and more than 22 million total users. Based on downloads, it is currently the t...

New post on Bluesky content moderation (cc: @noupside.bsky.social @nachristakis.bsky.social and @prossertj.bsky.social)

open.substack.com/pub/rajivset...

I’ve created my own starter pack!! It’s profoundly difficult to start as a cryptographer or engineer or programmer—then develop your presentation and political skills enough to explain tech well to policymakers/the public. Here are just a few of the people I learn from. go.bsky.app/8Ta3jza

I made a Black women in tech starter pack because I couldn’t find one and it annoyed me.

go.bsky.app/5upv7SQ

Careers at ACLU Join our team! We’re looking for committed, passionate people for open roles at the ACLU.

This looks like an amazing opportunity, and
@mkgerchick.bsky.social is a dream to work with. Worth applying if you're on the job market.

www.aclu.org/careers/appl...

reads: The Ways AI Decides How Low-Income People Work, Live, Learn, and Survive The use of artificial intelligence, or AI, by governments, landlords, employers, and other powerful private interests restricts the opportunities of low-income people in every basic aspect of life: at home, at work, in school, at government offices, and within families. AI technologies derive from a lineage of automation and algorithms that have been in use for decades with established patterns of harm to low-income communities. As such, now is a critical moment to take stock and correct course before AI of any level of technical sophistication becomes entrenched as a legitimate way to make key decisions about the people society marginalizes. Employing a broad definition of AI, this report represents the first known effort to comprehensively explain and quantify the reach of AI-based decision-making among low-income people in the United States. It establishes that essentially all 92 million low-income people in the U.S. states—everyone whose income is less than 200 percent of the federal poverty line—have some basic aspect of their lives decided by AI.

"92 million low-income people in the U.S. states—everyone whose income is less than 200 percent of the federal poverty line—have some basic aspect of their lives decided by AI"

www.techtonicjustice.org/reports/ines...

this is a damning report

Simulating human behavior with AI agents promises a testbed for policy and the social sciences. We interviewed 1,000 people for two hours each to create generative agents of them. These agents replicate their source individuals’ attitudes and behaviors. 🧵

arxiv.org/abs/2411.10109

I am recruiting 1-2 PhD students for Fall 2025 at Princeton to work on Comp Social Science/Societal Impact of GenAI/GenAI for SocSci

I wrote a bit on research flavor & interests here: manoelhortaribeiro.github.io/advising

Deadline: December 15th www.cs.princeton.edu/grad#prospec...

Please boost!

The AI Interdisciplinary Institute at the University of Maryland (AIM) is hiring

40 new faculty members

in all areas of AI, particularly:
- accessibility,
- sustainability,
- social justice, and
- learning;

building on computational, humanistic, or social scientific approaches to AI.

>

AI-Powered Buzzfeed Ads Suggest You Buy Hat of Man Who Died by Suicide Buzzfeed is using an AI-powered ad platform that recommends products that attempts to match objects in disturbing images on its site.

AI powered "shop this image" ads on Buzzfeed are helping people score the look of the astronauts who died in Challenger, a woman who needed amputations after a horrible infection, buy the hat of someone died by suicide, the look of an attempted murderer, etc

www.404media.co/ai-powered-b...

There’s No Longer Any Doubt That Hollywood Writing Is Powering AI Dialogue from these movies and TV shows has been used by companies such as Apple and Anthropic to train AI systems.

NEW: Today we published a searchable database of 139,000 movies and TV shows that have been used by companies such as Apple and Anthropic to train generative AI. Subtitle files taken en masse, giving huge amounts of natural dialogue to these programs. Gift link: www.theatlantic.com/technology/a...

Does the UK’s liver transplant matching algorithm systematically exclude younger patients? Seemingly minor technical decisions can have life-or-death effects

"It turns out that the algorithm is mostly just assessing need, that is, how long patients would survive without a transplant."

this is a damming essay about UK's liver transplant matching algo

www.aisnakeoil.com/p/does-the-u...

You can read the transparency reports here crfm.stanford.edu/fmti/May-202...

Starting to reup some of my work on here since I'm new to the Bluesky!

This year my team released the first transparency reports for large language models, helped win new protections for AI researchers, and presented at some of the top ML conferences.

A new @knightcolumbia.org blog argues for safe harbors for AI safety & research. By @shaynelongpre.bsky.social, @sayash.bsky.social, @klyman.bsky.social, @randomwalker.bsky.social, @epicfaace.bsky.social, @peterhenderson.bsky.social, Rishi Bommasami, Percy Liang. knightcolumbia.org/blog/a-safe-...

I am building a Bluesky feed for the internet policy community.

Add it by searching "Internet Policy" in the Feeds tab and send me suggestions or requests for folks I should add.