Hello fellow travelers,

Fasten your seatbelts because we will be passing through many travel metaphors and puns today.

Quarter three of 2025 carried us through newer terrains with AI Safety and Media Literacy. We also saw departures of team members charting new academic journeys and arrival of new copilots to keep us afloat.

Ready? Let's lift off together.

Viral Spiral Pilot in Mumbai

Viral Spiral is our multiplayer card game that you play with your friends to simulate social media behaviours and experience motivated reasoning and cognitive biases in action. You earn clout by sharing news with affinities and prejudices in them. The first person to get 10 clout before the world gets polarized wins.

Since all storytelling dreams come true in Mumbai, this September we went there to meet real board gamers and conduct guided playtests.

We met them all—cynics, enthu-cutlets and those who came "because my friend invited me."

One interesting feedback was that gamers did not want to read the card text and source text. This is very different from the media literacy and storytelling community, who actually enjoy reading those parts.

Gamers mainly wanted to compare card scores quickly so they could maximize points. This highlighted the challenge of balancing media literacy with fun gameplay.

We still strongly believe reading is an important part of the experience—we're definitely not contributing to the global decline in reading rates!

Follow us on Instagram for playtest updates.

Now that we're back from Mumbai, we're conducting live-stream playtests.

If you've already played Viral Spiral at least once, we'd love to have you join future playtests:

https://calendly.com/adhiraj-singh-r-11/30min

From Uli to AI Safety

The Uli project demonstrated that a participatory, expert-led approach was effective in building good datasets for online safety.

The same approach we used to crowdsource the slur list and the Uli dataset also helped us build LLM/AI Safety Benchmarks for MLCommons.

While many LLMs claim to be multilingual, their performance across languages varies significantly. Robust evaluations can reveal hidden vulnerabilities.

We're happy to be working with Tech4Dev on AI safety guardrails for nonprofits participating in their AI Cohort Program.

The cohort includes seven nonprofits building AI-powered solutions such as chatbots and information summarization systems.

We introduced them to frameworks such as the MIT AI Risk Repository and created a proof of concept demonstrating guardrails for chatbot applications.

Now Serving a Hot Take

Many people ask, "What even is open source AI?" as if the question itself ends the conversation.

We think there's actually a lot worth discussing.

Read our thoughts here:

https://tattle.co.in/blog/hot-take-on-open-source-AI/

Detours

We attended India FOSS once again.

The conference attracts many students exploring future career paths and is also an annual gathering of friends and like-minded organizations.

We hope we introduced attendees to alternative possibilities within open technology.

Our team presented talks on:

Fediverse
Open Source AI
Public Policy in FOSS

Work with Humane Intelligence

We also created an Open Source Release Plan for Humane Intelligence's AI red-teaming application.

This included:

Code review
Refactoring roadmap
Open source strategy
Governance recommendations
Licensing recommendations

https://humane-intelligence.org/product/oss-release/

Departures And New Arrivals

This quarter, two team members left Tattle to pursue higher education.

Saurav received the Chevening Award and will pursue an MSc in Health and International Development at the London School of Economics (LSE).
Aatman will pursue a Master's in Computational Linguistics at the University of Tübingen.

While we'll certainly miss both of them, we're also excited for their new journeys.

Thankfully, three new team members have joined us, and you'll hear more about their work soon.

Landing

We've arrived at our destination a little earlier than scheduled.

We hope you enjoyed the journey and will travel with us again next quarter.

In the coming months we'll be conducting workshops on:

Online Harassment (through Uli)
Media Literacy (through Viral Spiral)

If your organization or community would like to host one, we'd love to hear from you.

View this email in your browser:

https://mailchi.mp/dcb1377986ef/ai-safety-and-media-literacy-updates