Every trust & safety team faces the same impossible choices: Do you optimize for flexibility or performance? Accessibility or expressiveness? Speed or confidence?

Today is Safer Internet Day; a day for organizations, companies, schools, and communities to come together for a better internet. In the past year, I've come to believe that "together" can mean something different and powerful: building technology and making it free, easy to deploy, and in the open.

First community labeler taking actions on atproto using @roost.tools’s Osprey. entire Ozone and Osprey stack running on a $50/m OVH machine, with up to seven days of full firehose backfill for investigating patterns and exploring the network.

Osprey is a testament to how much impact we can create by sharing technology. @discord.com open sourced their internal rules engine through @roost.tools last year, working with us to generalize the core components so others can easily integrate it

How do you protect 39M users from AI-driven abuse? We’ve just published a deep dive into how @bsky.app uses Osprey to handle the scale of the AT Protocol firehose. roost.tools/blog/how-blu...

if you caught the @roost.tools talk at #FOSDEM and want to explore Osprey more, here's a codespace with about 14k test events to play with! turbo-fiesta-p46xvxwp4vfwqv-5002.app.github.dev try: SuspiciousDisplayName == True or PostContainsSpamLink == True

I had a great time at the Christchurch Call Foundation × @roost.tools brunch talking about Open Source and Online Safety with folks from across the ecosystem! You all had great insights. I look forward to continuing conversations about how to keep platforms and users safe with open source solutions.

Why does open source safety infra matter? Because bad actors are already sharing tools & components. It’s time the defenders did, too. Ystrdy, we launched Osprey V1.0, and we're doing a live demo at FOSDEM. Let us know if you’ll be there and say hi to @annebdh.bsky.social , @cassidyjames.com!

come find the ROOSTers at #FOSDEM!

getting to work on osprey a bit was really fun, and super happy to have gotten it in at bluesky. osprey opens up so much opportunity for the atproto ecosystem when it comes to not just moderation but also analytics, custom feed generation, and more. really hopeful that people start building with it!

Meet Osprey V1.0, a new open source online safety tool designed to help platforms investigate and address their priority threats at scale, without sacrificing data privacy or performance. roost.tools/blog/introdu...

Your analyst's favorite tools are about to get an upgrade. We’re heading to @fosdem.org to pull back the curtain on what we’ve been building, but we’re sharing the news here first with our community on Thursday. January 29, 2026 > Stay tuned.

that’s a lot of where my enthusiasm for @roost.tools comes from and why I do think the solution has to be both policy and tools, because if we’re able to close that trust&safety gap in a generalizable, reusable way for new entrants, it’ll be the first time that’s ever happened!

I am a big believer in the work of @roost.tools as a first-pass solution we can recommend to smaller apps for solving baseline moderation problems. I absolutely agree that most builders don’t have what they need out of the box right now!

The 2025 Social Web Trust & Safety Needs Assessment Report is now available. Volunteer moderators are burning out. Spam is rising. Legal threats are growing. The 2025 Report reveals where things stand, and what must change. ➡️ about.iftas.org/2026/01/08/t...

We just submitted to present at #ATmosphereConf 👀 We’re hoping to share about open source trust & safety infrastructure for distributed social media, including our Coop review dashboard and queuing tool! There are still a couple of says to submit your own proposal as well:

I only joined @roost.tools towards the tail end of 2025, but I’m proud of everything we’ve accomplished. In particular, I’m stoked that we open sourced Osprey, part of Discord’s mod tools, so that any platform can tackle T&S investigation and incident response in the same ways as the big platforms.

As we begin the new year, we’re taking a moment to reflect on how far ROOST has come since we officially launched at the Paris AI Action Summit. Ten months later, and we are entering 2026 with strong shared technical foundations and a growing community. roost.tools/blog/roost-i...

@roost.tools believes that digital & AI systems must be safe & accountable.✅ I wholeheartedly agree.💯 It’s an honour to join ROOST’s inaugural board of directors.🙏 Our tech, open source & resilience expertise will ensure we #FreeTheFuture🖖 — together! ▶️ roost.tools/blog/roost-w...

We’re proud to share an important milestone for ROOST: the formation of our inaugural Board of Directors! This Board brings together leaders across open source, technology, public policy, civil society, and national security. roost.tools/blog/roost-w...

Last week, something powerful happened. People from all around the trust and safety community came together. We hosted the Open Safeguard Hackathon in San Francisco with @hf.co and OpenAI —and it was a reminder of what’s possible when we build safety openly and together.

Along with our CTO, the TDC is co-chaired by Naren Koneru who leads engineering for Safety at Roblox. Prior to his time at Roblox, Naren spent a decade in open source software. His team is behind many of the new open source safety projects released in the last few years.

We're honored that Dr. Rebecca Portnoff is joining the TDC. Dr. Portnoff is an expert on AI and child safety and responsible and ethical AI systems. She has published many papers about online child safety and emerging harms, including the risks around AI generated CSAM.

Another fantastic TDC member is Sam Toizer, who leads the Safety Product team at OpenAI that's responsible for product safeguards at the model, system, and enforcement levels. Prior to OpenAI, Sam worked at Twitter for eight years working on platform integrity.

Joining the TDC is Shu Lei, currently engineering lead for Cash App's server team. Shu built a career at the intersection of innovation, safety, and global impact and spent the last 10 years leading Google Ads and YouTube Trust & Safety solutions. We're very excited for her expertise!

We have an open source development legend joining the TDC as well: @tpepper.bsky.social! Tim brings 30 years of experience all across the software stack and we're incredibly grateful he is spending some time making sure our projects are quality OSS projects

Also joining the TDC is @tgthorley.bsky.social, who joins with a background in user safety and countering harassment & malware detection, and eradicating CSAM and TVEC on GitHub. Prior to GitHub, Tom served at @gifct.bsky.social and spent 10 yrs at the British signals intelligence agency, GCHQ

And a big welcome to the inaugural Technical Design Committee. These are the community leaders shepherding @roost.tools projects! @hailey.at @tpepper.bsky.social @tgthorley.bsky.social Naren Koneru Dr Rebecca Portnoff Sam Tozier Shu Lei Vinay Rao

We are also very excited that @hailey.at is part of our Technical Design Committee! The initial TDC was selected by ROOSTers, but we aim to hold elections in the future. Along with community feedback and ROOST staff priorities, the TDC will help advise on our projects and priorities.

really happy and impressed with the momentum ROOST is building around open tooling and collaboration. particularly Osprey and Coop as bread-and-butter systems for human analysts and mod teams

@roost.tools isn't just openly licensing our tools, but also openly developing them. We just released an initial draft of the ROOST tooling roadmap, but it's just that -- a draft. If you're interested in OSS trust and safety tools, now's the time to shape what gets built! github.com/orgs/roostor...

We have just published our initial roadmap and announced our inaugural Technical Design Committee. We want to hear from all of you if we're focusing on the right things and what you think of our roadmap! Drop a comment in the GitHub discussion linked in the blog. roost.tools/blog/introdu...

This is Diffie. He’d like to be your coworker. Paws are no good for filling out job applications, though, so YOU’LL have to apply to the open software engineering roles at @roost.tools roost.tools/staff-softwa... (don’t fit the level? Apply anyways to express your interest in upcoming openings)

Hope you didn't miss out 👀 Digital safety nonprofit 🤝 unintentional self defense keychain for physical safety

On Monday, @hf.co , OpenAI, and ROOST hosted the first Open Safeguard Hackathon. The room was filled with people who care deeply about safety, working together to explore and test gpt-oss-safeguard. Check out the ROOST Model Community for some of the projects github.com/roostorg/mod...

This week, our President, @camillefrancois.bsky.social, visited the Vatican for a discussion on protecting children in the digital age. We were very humbled to have been invited to participate in such an important meeting and at such a pivotal time.

During our @parispeaceforum.bsky.social panel last week, Dr. Giada Pistilli eloquently explained that transparency in AI isn’t just a technical challenge — it’s a societal one. The more people who can see, shape, and question these systems, the more accountable and representative they become.

in case you missed it, gpt-oss-safeguard also came with a technical report that goes into the details of - limitations - multilingual performance - a new, more complex benchmark dataset of multi-turn conversation-style data PDF: cdn.openai.com/pdf/08b7dee4...

Thank you to Martin Tisné, Chair of the Board at Current AI and CEO of the AI Collaborative, for speaking on our panel this week at the Paris Peace Forum and for your contributions towards this work in open source and AI.

The @roost.tools delivered a landmark accomplishment in the Open Source ecosystem with the release of gpt-oss-safeguard with OpenAI. https://www.maffulli.net/2025/10/31/roost-foundation-releases-ai-safety-tool-with-openai/ #ai #safety

Thank you to Fidji Simo for addressing the audience via video yesterday ahead of our panel at the Paris Peace Forum. The conversation on how AI systems can help create a safer internet for the public good through open source tooling is only just beginning. lnkd.in/e_jkiemB

Here at ROOST, we have a saying. "Less words. More code."

At the Paris Peace Forum earlier today, we announced the release of gpt-oss-safeguard in partnership with OpenAI. Now, we’re announcing the launch of the ROOST Model Community (RMC). roost.tools/blog/a-new-milestone-for-open-source-safety-infrastructure-and-transparency/

Join @roost.tools, @hf.co, and OpenAI for a hackathon around gpt-oss-safeguard on December 8, 2025 in San Francisco! If you’re a developer, T&S expert, or policy wonk, see you there! events.openai.com/gpt-oss-safe...

We're excited to announce the result of months of collaboration with OpenAI, gpt-oss-safeguard. This reasoning model is designed for safety workflows where you bring your OWN policy and definitions of safety, and it explains each decision it makes. Learn more here: roost.tools/blog/a-new-m...

There's tried and true ways for how OSS applies to software, but that doesn't fit squarely with the realities of AI development. That's why I'm grateful OpenAI is trying something new with @roost.tools, openly releasing weights with a commitment to work with the ROOST Model Community to iterate

ROOST.tools has been building against the DIRE framework for online trust and safety: Detection, Investigation, Review and Enforcement. ROOST's Coop and Osprey tools address I, R, and E, but Detection is best addressed by classification models. Enter gpt-oss-safeguard from OpenAI

i'm also rly proud of writing up the User Guide for this T&S reasoning model: cookbook.openai.com/articles/gpt... It includes how this could work with @roost.tools first public tool, Osprey! cookbook.openai.com/articles/gpt...

this model will be available to use locally through @ollamabot.bsky.social ollama.com/library/gpt-...

The internet just got safer.🛜 Bravo @roost.tools & OpenAI for gpt-oss-safeguard!🙏 This open-source upgrade is a shared resource for the entire online safety commons.🫂 Co-creation & transparency will ensure we #FreeTheFuture — together.🤝 Read on roost.tools/blog/a-new-m... #LLAP🖖