OpenAI: We’re burning money like the Joker. A miracle needs to happen for us to turn a profit
Microsoft: Please please use our AI systems, we’re teetering on the edge here
Anthropic: I wonder what’ll kill us first: lawsuits, regulations or model collapse
Media and universities: AI is here to stay
close enough
just caught @brandontlocke.bsky.social using his motherfucking nose to scroll 👃🏻🤳🏻
6 pictures of old people in marine uniforms. 4 out of the six are screaming, the other two seem to show the same dude who looks quietly scared but isn't screaming
lately the chumbucket ads I've been seeing on local news pages keep featuring (ai generated?) stock images of old dudes in marine uniforms, usually screaming
NEW: The government tells Americans to use VPNs to protect their privacy. The government also automatically presumes communications of unknown origin are foreign. Foreign communications don't require a warrant to wiretap.
Lawmakers who've done the math want answers.
My latest @wired.com:
genuinely fascinating that we’ve hit a point where the full monetization of the internet and ai bot takeover has led to a majority of people wanting tech minimalism
so bummed i had to leave before this talk! thanks for a great weekend, #ATmosphereConf 💗
The BlackSky mod page
A restorative justice framework is intrinsic to @moderation.blacksky.app policy because there are always opportunities to change behavior and remain in community. It’s not about excluding people but about bounding acceptable behavior.
@drkalyncoghill.blacksky.team
that last point ("openness is taboo") is a persistent thorn in my side... motivated actors will always find ways to reverse-engineer and evade, even if you reveal nothing about your governance system. transparency is foundational for legitimate (i.e., successful) governance! #ATmosphereConf
two presenters in front of a slide, titled "Status quo: T&S is kind of broken" reasons include unaffordable vendor tools; platforms reinventing basic functionality; generative AI accelerating harms; openness being taboo
trust & safety status quo, per @julietshen.bsky.social and @cassidyjames.com at #ATmosphereConf: kinda broken! (to say the least)
Enjoy this detailed thread Juliet Shen kindly shared on @drkalyncoghill.blacksky.team 's #atmosphereconf talk on "Creating a Safer Web: Blacksky's Moderation Tool"
"There is no thing as a single-issue struggle because we do not live single-issue lives." -Audre Lorde
#BlackskyNews
these tools are what we make of them. we can't assume that because we are acting in benign ways, that everyone using the tools will do the same. People can't rely on only that. We need to discuss as a community how we govern labelers and community moderation. @scoiattolo.mountainherder.xyz
Blacksky's moderation tools grew from 2 labels and added new ones through a democratic process where people voted on new labels via their People's Assembly. Labels constantly get updated because things change and T&S needs to always have a finger on the pulse @drkalyncoghill.blacksky.team
it's today! come learn about coop, our review tool. we'll have a hosted demo up and running soon that everyone can play with too!
Before you begin How are you feeling right now こ Session Complete Acorn.Blacksky.Community
I think it’s very smart that @moderation.blacksky.app has built a process and tools (with Acorn) that, among other features, require that moderators take breaks and protect themselves. @drkalyncoghill.blacksky.team #ATmosphereConf
Where things go wrong in Trust & Safety: • Lack of community trust • Ignorance of cultural nuances • Unwillingness to acknowledge mistakes • Being reactive instead of proactive • Words + Actions that don't match = DANGER ZONE /! • Not understanding the community you serve
Key ways that people stop trusting Trust and Safety, per @drkalyncoghill.blacksky.team and @moderation.blacksky.app #ATmosphereConf
on blacksky, @drkalyncoghill.blacksky.team explains users aren't immediately banned—there's a conversation (plus opportunity to appeal), a less punitive approach informed by restorative justice. love to see restorative and transformative justice in the discourse at #ATmosphereConf 💗
reasons trust & safety teams aren't trusted, per @drkalyncoghill.blacksky.team at #ATmosphereConf:
- ignorance of cultural nuance
- unwillingness to acknowledge mistakes
- being proactive instead of reactive
- words that don't match actions
- not understanding the community you serve
💯💯💯💯💯
on blacksky, moderation is produced by actual community members (with considerations for their wellness), not exploited workers in the philippines #ATmosphereConf
love @drkalyncoghill.blacksky.team and blacksky's focus on explainability—ensuring moderated users understand exactly why their content was removed #ATmosphereConf
per @drkalyncoghill.blacksky.team, if you don't have your community's trust, your trust & safety practice needs work—regardless of your team's credentials (couldn't agree more!) #ATmosphereConf
if you enjoyed my #ATmosphereConf talk, make your way back to room 2301 for @drkalyncoghill.blacksky.team's talk on blacksky's moderation practices! from what i've been hearing so far, it sounds like they are doing everything right 💗
Devin speaking
@devingaffney.com’s work is hands down the best. @graze.social
Speaking truth to power!
🥲💞
if you missed the talk (and don't want to wade through an entire dissertation), "content moderation futures" is available here: arxiv.org/abs/2509.09076 #ATmosphereConf
highly recommend @julietshen.bsky.social for all your conference talk live-tweeting (and general cheerleading) needs
I'M SORRY i promise it's on the to-do list 😭
🫶🏻
So great to hear about @linguangst.bsky.social's moderation research - real clarity about the matrix of constraints and challenges that makes the work so difficult (and where we need to put in the effort to make it better) #atmosphereconf
