TechdirtCtrl-Alt-Speech: Honey, I Shrunk the Kids’ Internet
ROOSTWe’re back from #ATmosphereConf 2026 and we’re still thinking about it. @cassidy and @julietshen gave a trust & safety crash course, introduced Coop to the atproto community, and hosted an unconference session on what a minimum safety stack looks like for indie devs.
85+ people joined our talk live and via stream. We loved the conversations that followed; they’re exactly why we do this.
Cassidy James 
I wrote up the @roost experience at AtmosphereConf. Whether or not you’re interested in atproto, the conference was an incredible experience with one of the most welcoming and friendly communities I have ever had the chance to hang out with. The opening keynote alone should be watched by anyone building on the Internet.
https://roost.tools/blog/roost-at-atmosphereconf-2026/
There are a bunch of other freely available talk recordings linked in the post as well!
Ctrl-Alt-Speech: Age Old Questions
https://fed.brid.gy/r/https://www.techdirt.com/2026/04/02/ctrl-alt-speech-age-old-questions/
ZwillGenThe Fourth Circuit’s decision in United States v. Lowers offers an important reminder for providers navigating Trust & Safety and law enforcement requests, clarifying that provider access under a platform’s terms of service does not automatically extend to government access, and strengthens support for a broad, practical approach to reporting suspected CSAM. #TrustAndSafety
David Fleetwood - RG AdminPosts like this are why #IFTAS is not a trustworthy source of fedi moderation. Olifant was the instance admin who's moderators coordinated an attack on a trans leftist, slandering her repeatedly until she left the instance and then continuing to claim it was a 'both sides' situation.
IFTAS was formed soon after, but continued to work with people like Oli, take advisement from abusers and bring on people who defended the attack (including the cc in this message).
Tim GreenBalancing transparency in AI content moderation is a complex challenge: explaining decisions helps accountability but can aid bad actors. As regulations tighten, platforms must find ways to explain at scale without compromising security.
Discover more at https://smarterarticles.co.uk/the-transparency-trap-when-explaining-ai-moderation-helps-bad-actors?pk_campaign=rss-feed
#HumanInTheLoop #AIinContentModeration #RegulatoryTech #TrustAndSafety
Discover more at https://smarterarticles.co.uk/the-transparency-trap-when-explaining-ai-moderation-helps-bad-actors?pk_campaign=rss-feed
#HumanInTheLoop #AIinContentModeration #RegulatoryTech #TrustAndSafety
We are still hiring engineers, with a bit more well-scoped job descriptions now. Come work with us on open source trust and safety tools. 👀
We are a global, remote organization so let us know if you’re interested and have relevant experience!
Ctrl-Alt-Speech: For Meta Or Worse
https://fed.brid.gy/r/https://www.techdirt.com/2026/03/26/ctrl-alt-speech-for-meta-or-worse/
Ryan WinklerDomain Bugs Cost More Than Code Bugs
Domain-Driven Design, ubiquitous language, and bounded contexts matter because product teams ship the wrong workflow when legal, product, and engineering mean different things by the same word.
