Sign in

Juliet Shen

@julietshen.online
3.6K followers 2.9K following 3.7K posts

product @roost.tools. just another extremely earnest and Very Online idealist. long time Trust & Safety person full of personal opinions like the skeets below julietshen.online

PostsRepliesMedia
Juliet Shen @julietshen.online · 2h
I have been sick for 8000 years
050
Reposted by Juliet Shen
Vaishnavi @vaishnavi.bsky.social · 27/09/2026
Oh there’s been so much! I’m intrigued by this CSAM-auditing approach for large models without having to directly prompt them for CSAM news.mit.edu/2026/new-met...
news.mit.edu
New method aims to keep kids safe from illegal AI-generated content
Researchers developed an evaluation procedure that tests generative AI models for harmful capabilities without generating outputs. This could enable auditors to identify open-source models that have b...
141
Reposted by Juliet Shen
JD (by way of Blacksky) @jdot.blacksky.team · 29/09/2026
My new blog post looks at how adult content shows up on Blacksky and the options we're considering to make it a more consensual experience. We'd love for every community member to read it, think it over, and share their perspective. Let's build a bridge between self-expression and consent together.
whatshappeningonmyfeed.leaflet.pub
Your Body, Your Feed – Creating a bridge between consent and self-expression
Recently, we’ve seen an increase in discussion around how adult content currently shows up on our platform. By adult content, we mean content that is usually labeled as pornography, non-sexual nudity, or sexually suggestive content. Labeling adult content (both text- and image-based) is an example of how we empower community members to share more of themselves with folks, without creating a non-consensual experience for other members that do not wish to engage with adult content.  To solve this, we want to present you with options and invite you, as a member of our community, to decide how adult content shows up on Blacksky.
26023
Juliet Shen @julietshen.online · 25/09/2026
sick and it feels like my brain is a tortoise that I am standing over, yelling "PLEASE, FASTER" at
270
Reposted by Juliet Shen
Bailey Townsend 🦀 @pds.dad · 25/09/2026
Feel that? That's flannel weather.
3442
Reposted by Juliet Shen
Cassidy James @cassidyjames.com · 25/09/2026
More @roost.tools docs site updates! We’re now serving project docs from the shiny new docs.roost.tools domain. Idk what to do with the bare domain yet, so it currently redirects to roost.tools, but projects get a nice URL like: docs.roost.tools/coop/latest More to come soon. 👀
docs.roost.tools
Welcome - Coop Documentation
User guide, development guide, API reference, deployment info, and integration details for Coop, the open source review and moderation tool from ROOST.
063
Reposted by Juliet Shen
Global Project Against Hate and Extremism @globalextremism.bsky.social · 25/09/2026
Following the stabbing of Sikh driver Gurvinder Singh, GPAHE found a 66% rise in anti-Sikh & anti-South Asian posts across online platforms. Days before the attack, DHS posted an image telling “Mr. Singh” to “self-deport or find out.” Federal rhetoric is now echoing the hate targeting the community.
globalextremism.org
Hate Against Sikhs and South Asians is Surging Online. The Trump Administration isn't Helping
Anti-Sikh and anti-South Asian hate are spiking online and spilling into real-world violence. GPAHE found a 66% rise in hateful posts in the four days after Sikh truck driver Gurvinder Singh was stabb...
0179
Juliet Shen @julietshen.online · 24/09/2026
in the last 5 minutes of today: HAPPY BISEXUALITY VISIBILITY DAY took me forever to realize, but I'm so happy that I did
3291
Reposted by Juliet Shen
vortex_egg @vortexegg.com · 23/09/2026
One thing I appreciate from the (currently ongoing) Death of Online Anonymity online conference is the discussion about how LLMs (and widely accessible consumer AI generally) are power tools for deanonymizing people online. This is going to be a major problem for online harassment.
49735
Reposted by Juliet Shen
Laurens @laurenshof.online · 23/09/2026
AMSTERDAM LETS GOOO atmo.rsvp/p/did:plc:3x...
atmo.rsvp
AtmosphereConf 2027
The third edition of the global gathering for the Atmosphere Community is taking place in Amsterdam, April 29th - May 2nd. This includes side-events and main conference days. Please RSVP with your i...
2637
Reposted by Juliet Shen
rahaeli @rahaeli.bsky.social · 21/09/2026
There's also moderatormayhem.engine.is and the followup game trustandsafety.fun, which I like for showing some more of the real-world considerations but get less into detailed judgment calls
moderatormayhem.engine.is
Moderator Mayhem: A Content Moderation Game
Can you succeed at the fast-paced challenge of content moderation?
2123
Reposted by Juliet Shen
Aaron Rodericks @aaron.bsky.team · 21/09/2026
Compliments of @rahaeli.bsky.social - novehiclesinthepark.com is a good starting point for content moderation and trying to make calls with a standard set of rules when applied to different language. Will try to share more of these games so folks can practically learn how to make some calls!
novehiclesinthepark.com
No Vehicles In The Park
4254
Reposted by Juliet Shen
Vaishnavi @vaishnavi.bsky.social · 22/09/2026
Twenty countries now restrict children's access to social media. This fortnight: the EU KIDS Act, Meta's 332 CSAM, hundreds of Russian teenagers charged over online-recruited sabotage, and Kenya is building its evidence base.
quire.substack.com
EU Publishes KIDS Act With Tiered Age Limits, Meta Ran Hundreds of AI-CSAM Ads, & Russia Charges Hundreds of Teenagers Over Online-Recruited Sabotage
Digital Childhood Briefing, by Vys - September 21, 2026
011
Reposted by Juliet Shen
Riana @riana.bsky.social · 22/09/2026
Our top-line finding: existing laws are working just fine in AI-CSAM prosecutions. Most defendants pled guilty (as usual); the few who went to trial all got convicted. Prison sentences are longer than average for CSAM or obscenity cases. Out of 71 cases studied, only half a dozen are still pending.
1217
Reposted by Juliet Shen
Riana @riana.bsky.social · 22/09/2026
We’re also releasing the dataset of cases (with an appendix of 2026 cases). Our hope is that this data will be useful to others. I believe this is the largest public collection of AI-CSAM cases to date (though I assume there’s an internal list at DOJ). github.com/rspstanford/...
github.com
GitHub - rspstanford/aicsamdataset: Dataset of AI-related CSAM cases in federal court since 2022.
Dataset of AI-related CSAM cases in federal court since 2022. - rspstanford/aicsamdataset
1165
Reposted by Juliet Shen
Riana @riana.bsky.social · 22/09/2026
NEW RESEARCH PAPER! With Cyndie Demeocq at U of Edinburgh, today I’m releasing a paper analyzing 71 federal criminal cases, filed around the US from 2022-2025, that somehow involve AI-generated or -morphed CSAM or child obscenity. Here's the SSRN link: papers.ssrn.com/sol3/papers....
papers.ssrn.com
Yes, It's Already Illegal: Federal Criminal Prosecutions for AI-Generated Child Sexual Abuse Material, 2022-2025
<div> Generative artificial intelligence (AI) is now capable of creating realistic-looking images and videos with just a text prompt. This capacity has been mi
48136
Reposted by Juliet Shen
Raccoons Hourly @raccoonhourly.bsky.social · 22/09/2026
An adorable raccoon.
213328
Juliet Shen @julietshen.online · 22/09/2026
i'm streaming #UNGA from PBS on tiktok (lightly cursed, i know) and the comments are just a stream of bot comments about recognizing Somaliland and other bots saying it's part of Somalia
020
Juliet Shen @julietshen.online · 22/09/2026
"we must come together to build an open source frontier model. our investment capabilities must be pooled to build a cutting edge open source model. This should be shared and we cannot depend on somebody else." - Macron in his last address as president #UNGA
050
Juliet Shen @julietshen.online · 22/09/2026
macron talking about open weight AI capabilities at UNGA! let's goooo france always pushing public interest technology
120
Reposted by Juliet Shen
Rosemary Mosco @rosemarymosco.com · 22/09/2026
Happy mushroom season!
A comic titled "Types of mushrooms" with 12 mushroom species. A wrinkly-looking morel is labeled "creepy but tasty". A brownish deadly galerina is labeled "boring but deadly". A witch's butter on a branch is labeled "jiggles if you poke it". A large puffball is labeled "egg". A creepy red-dripping bleeding tooth fungus is labeled "deeply disconcerting". A fly agaric with a green treefrog sitting under it is labeled "frog umbrella". A black chanterelle is labeled "chalice for forbidden elixers". Three straight-stemmed entolomas are labeled "haha, nipples". A bright purple viscid violet cort is labeled "purple is nice". A birch polypore sticking out of a birch trunk is labeled "tree nose". A complex yellow-tpped coral fungus is labeled "cursed citadel where dwells a fairy queen". A Ravenel's stinkhorn is pixellated at the top for your safety, and labeled "NSFW".
5140311321
Juliet Shen @julietshen.online · 22/09/2026
I've turned my skincare / girls night group chat into an UNGA discussion chat
010
Juliet Shen @julietshen.online · 22/09/2026
my 9 animes to know me my9animes.com/en fighting foodons was bonkers but definitely fueled my cooking / food interests honorable mentions: digimon, neon genesis evangelion, pluto, paprika
an image of 9 anime covers: top row features Fruits Basket, Yu Yu Hakusho, Sailor Moon. The second row has Fighting Foodons, His and Her Circumstances, Junji Ito. Third row includes Bleach, Slam Dunk, and Tenchi Muyo
390
Juliet Shen @julietshen.online · 22/09/2026
daily appreciation post for @cassidyjames.com
static.klipy.com
Han Solo Star Wars
Alt: Han Solo Star Wars saluting
160
Reposted by Juliet Shen
Cassidy James @cassidyjames.com · 22/09/2026
The @roost.tools Coop docs are looking preeeeeetty nice now, huh? 👀 roostorg.github.io/coop/latest/
Screenshot of the Coop documentation with a ROOST theme. There is a nice Coop logo in the dark header, and a charcoal-colored sidebar navigation with the current page title in a bright yellow.The same documentation, but on mobile.
2103
Reposted by Juliet Shen
ROOST @roost.tools · 21/09/2026
If all goes well, we hope to release the stable 1.1.0 version later this week. 🤞️ Feel free to share any feedback here in this thread, or on this GitHub Discussion:
github.com
Coop 1.1.0-alpha pre-release feedback · roostorg coop · Discussion #1280
The community of ROOST contributors has been busy over the past several weeks working on the upcoming 1.1.0 release of Coop. We have a pre-release available for testing, and would love your feedbac...
041
Reposted by Juliet Shen
ROOST @roost.tools · 21/09/2026
The community of ROOST contributors has been busy over the past several weeks working on the upcoming 1.1.0 release of Coop. We have a pre-release available for testing, and would love your feedback if you’re currently deploying Coop or are interested in trying it out.
github.com
Release Coop 1.1.0 (alpha) · roostorg/coop
NoteThis is a pre-release! We would appreciate early testing and feedback of these updates to help us validate the stable 1.1.0 release. The world’s first free, open source child safety infrastruc...
185
Reposted by Juliet Shen
Emelia @thisismissem.social · 22/09/2026
Wooo, first Coop release I've directly been a part of — hoping you'll see more of that soon! I've been grinding on reducing tech debt and figuring out how to make the code more maintainable long term
2464
Juliet Shen @julietshen.online · 22/09/2026
QUEEN EMILIA
190
Reposted by Juliet Shen
Cassidy James @cassidyjames.com · 22/09/2026
The @roost.tools community site now shows contributors from the GitHub API! It doesn’t reflect every contribution to the community (e.g. folks who are active in working groups or chat but not on GitHub), but it’s a start to try to visualize who the ROOST community actually is. community.roost.tools
Screenshot of the ROOST community site. It reads “Meet us: These are the people building ROOST projects in the open.” Below the text are 80+ GitHub avatars of contributors.
1103
Reposted by Juliet Shen
Torsten Goerke (verified 🤜 🤛) 🎛️🎹🔊🎚️🎧📟🔌🐸 @tgoerke.bsky.social · 22/09/2026
Also this post by @julietshen.online on Jev for T&S sparked my interest to use it for judgement / decision tasks. I am now toying around with the #ATScience community labeller idea. Sharing my moodboard for "Atmospheric Evidence Signals" @breadcrumbs.run docs.breadcrumbs.run/artifacts/at...
docs.breadcrumbs.run
Atmospheric Evidence Trail
031
Reposted by Juliet Shen
LenapeDevil @work2ski.bsky.social · 22/09/2026
Tonight, we see what Sunny Mehta has built, one of the deepest top-nine forward groups in the NHL on the ice. May tonight be the start of something special. #NJDevils
021
Reposted by Juliet Shen
Jersey boy in 'Straya @rogersherman.com · 22/09/2026
Things I hope for, this season: Healthy Hugheses Beating the Canes A nice playoff run My Canucks fan friends to have some joy this season Reddit to burn to the ground. #NJDevils
181
Juliet Shen @julietshen.online · 22/09/2026
NICO FOREVER OUR KING #NJdevils
020
Juliet Shen @julietshen.online · 22/09/2026
the benefit of working with many French people is that I learned that I've been saying a very vulgar word when pronouncing Mistral's "Le Chat" for the past 2 years
0220
Juliet Shen @julietshen.online · 21/09/2026
some of the editing I've been doing for @wikipedia.org got covered by in Everything in Moderation, a newsletter by @benwhitelaw.bsky.social! There is still a lot to do, and if there's one thing wiki and atproto have taught me: WE CAN JUST DO THINGS! www.everythinginmoderation.co/trust-safety...
everythinginmoderation.co
Why T&S needs a Wikipedia-thon
Wikipedia is often the first stop for anyone trying to understand T&S and yet its coverage of the industry is woefully thin. It's time that was fixed
4344
Juliet Shen @julietshen.online · 21/09/2026
fortune weave just makes me miss the story rich supports and characterization of three houses I guess I'll replay 3H for the 13th time
020
Juliet Shen @julietshen.online · 18/09/2026
shout out to @annebdh.bsky.social for the book recommendation, going to start soon! cup.columbia.edu/book/the-lan...
cup.columbia.edu
The Language of Terrorists | Columbia University Press
If someone posts a death threat or a violent manifesto online, will they cross over from words to deeds? Explicit threats, intimidation campaigns, and incite... | CUP
271
Reposted by Juliet Shen
Cassidy James @cassidyjames.com · 15/09/2026
It’s certainly an interesting way to literally overhear people’s public thoughts while learning about something/someone. Pictured here: an exhibit about Macklemore. There are lots of thoughts!
121
Juliet Shen @julietshen.online · 18/09/2026
this is so awesome
040
Reposted by Juliet Shen
Cassidy James @cassidyjames.com · 11/09/2026
Okay hear me out: combine this with The Museum of All Things from @may.as, where you explore an infinite museum that is procedurally generated from the Wikipedia API. Spawn the Throng people in exhibits related to what they’re talking about.
mayeclair.itch.io
Museum of All Things by Maya Claire
An infinite virtual museum generated from wikipedia
231
Reposted by Juliet Shen
Streamplace @stream.place · 14/08/2026
And we have basic machine-vision in place to help identify harmful content. We understand discomfort with these systems, but there's great potential for harm in live video; machine-assisted moderation has long been considered a T&S best practice. Humans will always make the final rulings on bans.
1542
Reposted by Juliet Shen
ROOST @roost.tools · 17/09/2026
Most tooling for fighting fraud and abuse was built for 2010s social media. The abuse evolved. The tools mostly didn't. @julietshen.online at Marketplace Risk NYC on why open source and open weight models help: you run them yourself, so sensitive data never leaves your systems. github.com/roostorg
ROOST Head of Product Juliet Shen speaking at Marketplace Risk in NYC.
0121
Juliet Shen @julietshen.online · 17/09/2026
I was at this show last night and I noticed a ton of Macklemore, and it's awesome to see this confirmed! I love her sm stereogum.com/2511718/hayl...
stereogum.com
Hayley Williams Playlist Features Acts Who Left Ed Sheeran Tour
A week ago, it might’ve been surprising to go see Hayley Williams and to hear Macklemore and Lukas Graham on the pre-show playlist. Now, it looks like a righteous statement of solidarity. Late last we...
2130
Reposted by Juliet Shen
Daphne Keller @daphnek.bsky.social · 17/09/2026
(1) This is amazing 👏 (2) The better automated content moderation gets, the more reasonable it is for govts to mandate automated censorship 🤐 (2) Having smaller/cheaper/more customizable como means there’s less reason to have centralized speech controls in the first place 🤓
074
Juliet Shen @julietshen.online · 17/09/2026
alright Nico, Jack, and Jesper! let's make it a good season boys 👹
020
Juliet Shen @julietshen.online · 16/09/2026
has anyone played around with jev for safety or moderation use cases yet?
3100
Juliet Shen @julietshen.online · 16/09/2026
FORTUNE WEAVE COMES OUT TOMORROW FORTUNE WEAVE COMES OUT TOMORROW FORTUNE WEAVE COMES OUT TOMORROW FORTUNE WEAVE COMES OUT TOMORROW FORTUNE WEAVE COMES OUT TOMORROW FORTUNE WEAVE COMES OUT TOMORROW FORTUNE WEAVE COMES OUT TOMORROW FORTUNE WEAVE COMES OUT TOMORROW FORTUNE WEAVE COMES OUT TOMORROW
010
Juliet Shen @julietshen.online · 16/09/2026
My talk for the Trust & Safety Professional Association APAC Summit was accepted! I'll be presenting on "Open Safety Models and the New T&S Tooling Stack" I'm poking around at Asian language support and performance in bring-your-own-policy models and fine tuned ones: github.com/julietshen/v...
github.com
vibecheck/MULTILINGUAL.md at main · julietshen/vibecheck
Playground for evaluating various safety tools. Contribute to julietshen/vibecheck development by creating an account on GitHub.
2170
Reposted by Juliet Shen
U.S. Consumer Product Safety Commission @cpsc.gov · 16/09/2026
Happy Baby Safety Month.
833612916