Big takeaway from StarPal AI Summit: real-world AI success is less about model power and more about trust + safety by design. Clear data rules, output checks, human oversight, and fail-safes. That’s how we build at Guildford.AI. #AI #TrustAndSafety #AIGovernance

Alecto Core | Explainable Content Moderation API and Integrations Content moderation infrastructure for chat, forums, marketplaces, communities, and support workflows. Every outcome is auditable, explainable, and defensible.

Real-time chat moderation has one hard constraint: the user is waiting.

Add latency and your product feels broken. Block on failure and it is broken.

Fail open or closed. You control it.

alectocore.com

#ContentModeration #TrustAndSafety

I am once again asking for help from folks with experience with text classification pipelines and building effective scoring systems from them!!#atproto #machinelearning #trustandsafety

Fourth Circuit Clarifies Two Important Points for Providers in United States v. Lowers Lowers provides more clarity on providers’ Trust & Safety rights and responsibilities—both of which ultimately prioritize user safety.

The Fourth Circuit’s decision in United States v. Lowers clarifies that provider access under a platform’s terms of service does not automatically extend to government access, strengthens support for a broad approach to reporting suspected CSAM. #TrustAndSafety

www.zwillgen.com/law-enforcem...

One month left to submit your proposal to the 5th Annual #TSRConf!

Join leading researchers, practitioners, policymakers, and platform leaders shaping the future of online #TrustAndSafety and #DigitalGovernance.

📩 Submit your proposal and be part of the conversation!

🔗 https://bit.ly/4agUJ0f

Out now in the Spring 2026 issue of #JOTS, a new study by @kingcatherine.bsky.social, Samantha Phillips, and Kathleen Carley surveyed active American #SocialMedia users to understand what drives public support for #misinformation interventions.

#TrustAndSafety

Authored by Fatmaelzahraa Eltaher, Rahul Krishna Gajula, Luis Miralles-Pechuán, Patrick Crotty, Juan Martínez-Otero, Christina Thorpe, and Susan Mckeever

#JOTS #TrustAndSafety #SocialMedia #SocialMediaSafety

@leahf.bsky.social #AdityaGautam #ChrisMiles #OmriTubiana #ArushiSaxena #JJMartinezLayuno #DavidJay

#AI #LLMs #Misinformation #TrustAndSafety #Ethics #ResponsibleAI #TechPolicy #ContentModeration #Governance #DigitalTrust #PlatformAccountability

@aaron.bsky.team shouting out how great using Osprey is for investigating! Go @roost.tools @julietshen.bsky.social #opensource #tssummit #trustandsafety

Disinformation isn't just a geopolitics problem, it affects local democracy too. That's why this talk was organised by municipal council factions in Ulm and Neu-Ulm.

Read the full article (German, paywall): swp.de/lokales/ulm/...

#TrustAndSafety #AI #Disinformation 5/5

#conectys #businesstransformationoutsourcing #bto #cx #trustandsafety #ai | Conectys Introducing Conectys’ fresh look with a distinctly digital edge. ✨ New look, same drive, sharper focus. Same obsession with outcomes, driven by hard work, bold ideas, and big ambitions! Our new vis...

Conectys reveals new logo and brand identity.

#CX #CXServices #TrustandSafety

@nhinsight.bsky.social

www.linkedin.com/feed/update/...

Transparency Report Update: July – December 2025 The 25th edition of our biannual transparency report, covering the period from July through December 2025 is now available. The work of Automattic’s Trust & Safety team is grounded in key princ…

Although the DMCA and DSA are important tools, this report shows that they are not immune to misuse—particularly as bad actors increasingly weaponize AI to exploit them.

transparency.automattic.com/2026/02/23/t... #transparency #reports #TrustAndSafety

brinsa.com/synthetic-sw...

#AI #Deepfakes #RomanceScams #OnlineDating #Cybercrime #FraudPrevention #DigitalIdentity #Disinformation #TrustAndSafety #AIGovernance #SocialEngineering #ConsumerProtection #seikouri #chatbotsbehavingbadly

SHIELD Global Online Safety Conference SHIELD Global Online Safety Conference 2026. The Future of Tech, From the Ground Up

Excited to be speaking at the SHIELD Global Online Safety Conference.

We'll look at which users are rendered invisible by current safety standards and how we can course-correct.

Details here: shield-global-online-safety-conference.heysummit.com

#OnlineSafety #TrustAndSafety #DigitalPolicy

TaskUs Announces Fiscal Fourth Quarter and Full Year 2025 Results | TaskUs NEW BRAUNFELS, Texas --(BUSINESS WIRE)--Feb. 25, 2026-- TaskUs, Inc. (Nasdaq: TASK), a leading provider of outsourced digital services and next-generation customer experience to the world’s most innov...

TaskUs FY 2025 results: revs up 19.0% y/y to $1,183.5m, net income margin up 4pts y/y to 8.6%. FY 2026 outlook: revs between $1,210m to $1,240m. Announces CFO transition.

#CXServices #CX #TrustandSafety @nhinsight.bsky.social

ir.taskus.com/news-release...

How Bluesky Transformed their Safety Operations with Osprey Robust Open Online Safety Tools or ROOST is a new non-profit entity designed to address the urgent need for accessible, high-quality safety tools in the rapidly evolving digital landscape.

And of course, @safety.bsky.app is also already using Osprey to handle tens of millions of daily events, tackling real-time social media issues like scams, spam, and more.

It’s an exciting time for open source trust and safety!

#OpenSource #TrustAndSafety #Bluesky

@matrix.org is doing just that by integrating Osprey w/policy servers for automated rules and investigations. Users of the open source, decentralized alternative platform will be safer, too, thanks in part to @roost.tools. 😉

#OpenSource #TrustAndSafety #Matrix

Osprey: Open Sourcing our Rule Engine Discord uses Osprey to quickly detect and remove new types of harm from putting our customers at risk. Now we’re open-sourcing this tool so others can do the same.

@discord.com open sourced their internal T&S rules engine with @roost.tools!

Osprey handles ~400 million actions per day in production at Discord. If you run a Discord-sized (or smaller!) platform, you can just… use their rules engine because it’s open source.

#OpenSource #TrustAndSafety #Discord

AI chatbots aren’t equally helpful. MIT tested GPT-4, Claude 3 Opus & Llama 3: accuracy fell for non-native English + less-educated users, and Claude refused ~11% of Qs vs 3.6%.

#AI #GenerativeAI #LLM #ChatGPT #Claude #Llama3 #AIEthics #AlgorithmicBias #TrustAndSafety #EdTech

📢 Last call! 📢

Submit your commentaries for the Journal of Online Trust and Safety by March 1 to be considered for the Spring 2026 #JOTS issue.

We invite letters, editorials, or other #TrustAndSafety research outputs to be submitted as commentaries.

Details & submission form ⤵️
bit.ly/4oi8b97

Coop v0 Released! · roostorg coop · Discussion #44 As you may have seen, we've released the initial public version of Coop, tagged as v0. The release includes core review capabilities alongside specialized child safety workflow functionality. There...

Today @roost.tools released v0 of Coop, the trust & safety review tool. 👀

Note that it’s v0 for a reason! While we focused on core functionality and child safety features like Google Content Safety API integration(!)…

#TrustAndSafety #OpenSource

Beyond Community Notes: A Framework for Understanding and Building Crowdsourced Context Systems for Social Media Social media platforms are increasingly adopting features that display crowdsourced context alongside posts, a technique pioneered by X's Community Notes. These systems -- which we term Crowdsourced C...

Paper: arxiv.org/abs/2509.15434
w/ @tungdnguyen.bsky.social @karenlevy.bsky.social @informor.bsky.social
#communitynotes #socialmedia #research #paper #HCI #socialcomputing #misinformation #trustandsafety #contentmoderation #crowdsourcing

ROOST at FOSDEM 2026 Showing up for the Open Source community

I’m back from FOSDEM! Here’s a little blog post about our first time attending as @roost.tools:

cassidyjames.com/blog/roost-o...

#TrustAndSafety #FOSDEM #FOSDEM2026 #OpenSource

No platform can tackle terrorist and violent extremist abuse online alone. @gifct.bsky.social brings platforms together with shared tools, expertise, and a trusted space to collaborate - helping turn collective action into real progress on online safety.

#GIFCT #TrustAndSafety #CVE

Let’s build due process! Part 2: A workflow that works This is the big one - my proposal to restore user trust in the account termination process without breaking the bank. Ask your favorite platform for this. Or if you work at a platform, maybe even give...

What do you think? If you work in trust & safety, would you consider building this? Why or why not? Read the full proposal at Platformocracy (or subscribe to get ideas like this in your inbox every Friday morning.) 7/7 #trustandsafety

RE: https://hachyderm.io/@thisismissem/116012963302634018

From one of the items on the agenda:

"How could peer moderation work in an ActivityPub context?"

#fediverse #moderation #TrustAndSafety

Presenter Application: Trust and Safety Research Conference 2026 The Stanford Tech Impact and Policy Center's two day Trust and Safety Research Conference focuses on research in trust and safety for those in academia, industry, civil society, and government. The co...

Apply by April 30 to present your work at #TSRConf through a presentation, lightning talk, poster, participant-organized panel, or workshop.

Don’t miss this opportunity to share your #TrustAndSafety research with a community dedicated to making the internet safer for everyone.

👉 bit.ly/4agUJ0f

📢 Call for proposals for #TSRConf 2026 📢

Mark your calendars! The Trust and Safety Research Conference returns on October 1–2, 2026, bringing together 500+ professionals from academia, industry, civil society, and government to tackle the most pressing questions in #TrustAndSafety research.

Me standing in front of the BXL sculpture

Good morning Brussels! I’m on my way to the campus to give a talk about how @roost.tools is bringing open source to trust & safety (and vice versa). We’ll be in AW at 10 AM!

fosdem.org/2026/schedul...

#FOSDEM #FOSDEM2026 #FOSDEM26 #TrustAndSafety