New paper in PNAS: Fact-checking information from large language models can decrease headline discernment What are the unintended consequences of AI fact-checking? Even LLMs that accurately identify false headlines can actually cause harm. doi.org/10.1073/pnas.2…

Check out our new work 👇

🚨Our new paper is on arXiv! We’re excited to share 26 state-backed information operation (IO) datasets, each paired with a robust control set of 13M+ posts in total from non-IO accounts. You can reach the dataset here: zenodo.org/records/141891… ...

So nice to catch up with Onur Varol at #ComplexNetworks2024 in beautiful Istanbul!

🌟 Our 2024 Annual Report is here! 🌟 We’re excited to share the highlights, achievements, and incredible work of our team at the Observatory on Social Media this year. Discover what we've been up to: osome.iu.edu/research/blog/…

The three horsemen of social media: brain rot, anxiety and foreign interference voxeurop.eu/en/social-medi…

Nice explanation about our recent paper in PNASNews w/Matthew DeVerna psypost.org/chatgpt-fact-c…

🔊 Don’t miss our next OSoMe Awesome Speaker: Dr. Deen Freelon! 📅 Feb 6, 12 PM ET | 💻 Online 📢 Computational Research in the Post-API Age 🔗 Register: iu.zoom.us/meeting/regist…

Don't forget our OSoMe Awesome Speaker Deen Freelon tomorrow!

🚨 Can crowdsourced moderation improve online discourse? 🚨 Join us March 5 at 12PM ET for a virtual talk with Dr. Amy Zhang on how platform design can empower users to moderate content & improve information quality at scale. 🔗 Register now: iu.zoom.us/meeting/regist…

New paper: Modeling the amplification of epidemic spread by individuals exposed to misinformation on social media Simulations informed by social media data yield a worst-case bound on additional infections due to exposure to online vaccine misinfo. nature.com/articles/s4426…

Maintenance is finally complete! Botometer X and the rest of our tools are functioning properly again! 🎉

Paper alert! 🚨 We investigate Bluesky from an invitation only platform with a few thousands of users to reaching 30 million users in terms of user activity and network. 📄 Full paper: arxiv.org/abs/2504.12902 💻 Codes: github.com/osome-iu/rise_… 💾 Dataset: zenodo.org/records/150660…

Bluesky's growth was driven by key events: the public access launch, Brazil's ban on X, X's block policy change, and the US elections. Each event triggered significant user migrations and formed distinct communities.

Despite initial bursts and declines in total and average activity, Bluesky achieved a stable level of daily engagement, with around 15% active users after the US Elections, matching inactivity rates on X/Twitter (~85% lurkers).

Our analysis reveals that Bluesky rapidly developed a dense, highly clustered network structure. This "friend-of-a-friend" connectivity, characterized by strong hubs, enables swift and viral information diffusion, similar to established platforms such as Twitter/X and Weibo.

As Bluesky continues to mature, influential accounts are emerging, posing familiar risks of misinformation, abuse and toxicity on platform. Understanding these dynamics can help inform effective governance and moderation strategies moving forward.

Note that a toxic post is defined as a post with >.5 toxicity score from OpenAI moderation endpoint. platform.openai.com/docs/guides/mo…

All is done by my wonderful co-authors here! Filipi Silva, Bao Truong, Sangyeon Kim, Fan Huang, Nick Liu, Alessandro Flammini, Filippo Menczer, Observatory on Social Media

Our team just launched three new tools to help you explore social media data: 🌉NewsBridge – Adds AI-powered context to Facebook posts 🔎Barney’s Tavern – Search 34B social media posts 🌎OSoMeNet – Visualize how info spreads across platforms osome.iu.edu/research/blog/…