Group size effects and collective misalignment in LLM multi-agent systems arxiv.org/abs/2510.22422

New preprint: More is different for AI multi-agent systems - but what does more mean? arxiv.org/abs/2510.22422 With @ariel-flint.bsky.social, @lajello.bsky.social and Romualdo Pastor-Satorras.

Not being an AI-doomer, but having experienced in my own department over the past 7 years a steady erosion of faculty governance norms and diminished prioritization of research-oriented pedagogy, this tracks. Let faculty and faculty interests lead the way, rather than administrators/regents

“LLMs don’t understand.” Of course. That was never the point. The point is: we’re already using them as if they do — to moderate, to classify, to prioritize, to decide. That’s not a model problem. It’s a systemic one. The shift from verification to plausibility is real. Welcome to Epistemia.

'I once met a former bodyguard of one billionaire with his own "bunker", who told me his security team's first priority, if this really did happen, would be to eliminate said boss and get in the bunker themselves. And he didn't seem to be joking.'

Researchers say Israeli government likely behind AI-generated disinfo campaign in Iran cyberscoop.com/citizen-lab-...

It was a ton of fun to play a small role in this project lead by the brilliant Marta Ewa Lech & @jonassjuul.bsky.social. A lot of drama & change in the world of music reflected in the song-dynamics of the Billboard Hot 100!

There are two ways to deal with LLMs in our work One is the path of abundance: automate everything and multiply publications at the expense of quality The other is the path of focusing on the soul of the work, on the questions that still matter - to you - when everything else has been made trivial

Breaking: the court ruling in Wikipedia's legal challenge to the Online Safety Act categorisation regime has just been handed down. The court dismissed Wikipedia's challenge. This is bad. This is very bad. wikimediafoundation.org/news/2025/08...

Presented our work, “Cultural and Cognitive Insights from 2.6 Billion Drawings,” at @ic2s2.bsky.social this morning. Big thanks to Mauro Martino, Nima Dehmamy, @douglasguilbeault.bsky.social, @lajello.bsky.social, and @baronca.bsky.social for the collaboration ✨ Stay tuned for more results!

Looking forward to it!

🚨Are moral judgements biased by the gender of the subject? We find *no* direct effect when looking at Reddit! Our causal study design is based on carefully matching similar situations that people with different genders experience (and share). Study led by @loreb92.bsky.social rdcu.be/eumlQ

Our preprint “How Malicious AI Swarms Can Threaten Democracy” is now online. We show how coordinated multi-agent LLM swarms can infiltrate communities, craft synthetic grassroots “consensus,” and erode institutional trust—and what we can do now to stop it. osf.io/preprints/os...

New paper hot off the press www.nature.com/articles/s41... We analysed over 40,000 computer vision papers from CVPR (the longest standing CV conf) & associated patents tracing pathways from research to application. We found that 90% of papers & 86% of downstream patents power surveillance 1/

At #ICWSM? Here's a starter pack for you :) bsky.app/starter-pack... Check also "Bootstrapping Social Networks: Lessons from Bluesky Starter Packs" ojs.aaai.org/index.php/IC...

A brief follow-up on the complex systems foundations of our LLM population study, sparked by ongoing debate. And why they matter for studying AI collectives more broadly. arxiv.org/abs/2506.18600 With @ariel-flint.bsky.social and @lajello.bsky.social

Our work on Bluesky: doi.org/10.1609/icws... Joint work with @bibo7086.bsky.social @harnen.bsky.social @garethtyson.bsky.social @asonur.bsky.social @ignactro.bsky.social @baronca.bsky.social

Average day in AI discourse: - The big tech paper saying LLMs are harmless toys, - The influencer claiming LLMs are an existential threat.

👇👇👇

Only two weeks left to apply!

Nice summary of our latest paper. www.nature.com/articles/d41...

New paper alert. "More is different" & the LLMs. LLMs are usually studied in isolation. But what happens when they start interacting? We explored this by looking at their collective behaviour. Work with @ariel-flint.bsky.social and @lajello.bsky.social 1/

Oral exams are the only true way to assess students - except in specific, marginal cases like parts of Maths or Physics. This has always been true, and we can’t ignore it anymore.

🚨 Just 1 week to go! Join us in London 🇬🇧 on May 16 for the NetPlace Workshop on Networks & Physics! Keynotes, collabathon, ERC-style pitches & more 🧠⚛️💥 🆓 Registration is free 💸 Travel/accommodation support available for those who register by Friday May 9 👉 netplace.site/event/netpla...

*Permanent* position in Applied Mathematics at City. We are - of course - particularly interested in profiles in Computational Social Science, Network Science, Data Science, and related fields. The application deadline is June 1st. www.jobs.ac.uk/job/DMY889/l...

New preprint on the global state of political polarization research 🎉 osf.io/preprints/os... We discuss the many different ways polarization can be conceptualised, how it manifests very differently across countries and domains, and the problems with the dominance of US focused research.

Richard Feynman: "My request for resignation from the National Academy of Sciences [..] flow[s] from the fact that most of the membership consider their installation as a significant honor."

Some people wonder why Musk was a member to begin with. Remember that this is Brittain. The society also has royal fellows: The King of the United Kingdom, The Duke of Kent, The Princess Royal, and The Prince of Wales. Makes perfect sense to also include the real masters of the universe.

“There’s a change in how willingly these individuals express their views--they’re certainly more visible.” @bertiehb.sky.social interviews @baronca.bsky.social and @maxfalken.bsky.social on the emergence of climate denial and scepticism on X. From the archive: 12/1/23 www.landclimate.org/...

#CulturalDataAnalytics Today! 🔥Andrea Baronchelli🔥 (@baronca.bsky.social) "From Local #Interactions to Global #Conventions: Emergent Social #Coordination in #Humans and #Machines." @cudanlab.bsky.social 16:00-18:00 UTC+2 Zoom & abstract: cudan.tlu.ee/events/2025-... #CompSocSci #GenAI #Bias

Influcast, Italy’s first collaborative epidemic forecasting hub for influenza-like illness. A multi-team forecasting hub that integrates 8 models from 5 teams to predict ILI incidence up to 4 weeks ahead, both nationally and regionally. Report of season 2023/24 www.sciencedirect.com/science/arti...

🎉 Today, our perspective on "Using mobility data to quantify experienced urban inequalities" is finally published in Nature Human Behavior. I am honored to be listed among many stellar coauthors and thankful for their valuable insights. www.nature.com/articles/s41...

Can simple closed-form mathematical models predict human mobility as well as deep learning? In a new paper in @naturecomms.bsky.social we show that the answer is YES Human mobility is well described by closed-form gravity-like models learned automatically from data www.nature.com/articles/s41...

EU to invest €200bn in AI, aiming for a collaborative ecosystem "akin to a CERN for AI." ec.europa.eu/commission/p...

We took X to court. And we won! ⚖️ The courts ruled that X must urgently grant us access to their publicly available data via their API. This access is critical for our analysis of online public discourse ahead of the upcoming German elections. Read: bit.ly/4gBWGp4 Here's what's happened 1/3 👇

Bad news for UK residents in today's The Washington Post www.washingtonpost.com/technology/2...

Bumping this again! IT:U is a great place to work. Groups are very well equipped, the culture is perfect for interdisciplinary research! You also get the opportunity help shape a brand new university. Deadline is Feb 28, 2025. Get in touch if you want to learn more.

The myopic obsession with impact is now devouring itself. No direct impact means death - even for maths, which has impact on *all* the most impactful fields. #UK #STEM www.ft.com/content/4ed6...

Bluesky has just surpassed 30M users. Starter packs have been a key feature - how have they worked out so far? Check out our recent preprint for insights: arxiv.org/abs/2501.11605

📢 𝗡𝗲𝘄 𝗽𝗮𝗽𝗲𝗿 alert! We document how 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 𝗶𝘀 𝗿𝗲𝘀𝗵𝗮𝗽𝗶𝗻𝗴 𝗱𝗲𝗺𝗮𝗻𝗱 𝗳𝗼𝗿 𝗳𝗿𝗲𝗲𝗹𝗮𝗻𝗰𝗲𝗿 𝘀𝗸𝗶𝗹𝗹𝘀. 💼 💻 📊 We analysed 𝟯𝗠+ 𝗷𝗼𝗯 𝗽𝗼𝘀𝘁𝘀, used embeddings to cluster them into 𝟭𝟬𝟬+ 𝗳𝗶𝗻𝗲-𝗴𝗿𝗮𝗶𝗻𝗲𝗱 𝘀𝗸𝗶𝗹𝗹𝘀, and LLMs to classify them as s𝘂𝗯𝘀𝘁𝗶𝘁𝘂𝘁𝗮𝗯𝗹𝗲, 𝗰𝗼𝗺𝗽𝗹𝗲𝗺𝗲𝗻𝘁𝗮𝗿𝘆, or 𝘂𝗻𝗮𝗳𝗳𝗲𝗰𝘁𝗲𝗱.

Periodic reminder (quite timely this time): using LLMs requires care, but banning them for writing discriminates against non-native English speakers and people with dyslexia - and that’s not acceptable.

And here we go! First position on my ERC grant is officially open. Fully funded 3.5 years PhD position at NetSI London working with me and the @nplab.bsky.social on modeling and reconstructing higher-order systems. Application deadline: 16th February 2025. Bring your A game!

I’m looking for a new PhD student with a curious and interdisciplinary mindset to work on integrating multimodal data to better understand trajectories of mental and brain health disorders! Deadline of applications: Feb 16! Please spread the word. www.jobs.ac.uk/job/DLO680/f...

"Bootstrapping Social Networks: Lessons from Bluesky Starter Packs" 👇👇👇

Executive Order 14110, "Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence", has been revoked by U.S. President Trump as of yesterday. www.whitehouse.gov/presidential...

Kicking off 2025 ANETI seminars with an exciting talk on emergent social coordination in humans and machines by @baronca.bsky.social If you are in Budapest, you can join us in person, otherwise you can attend online. More info here 👇 anetilab.org/seminars/

Announcing the 1st Satellite on Understudied Networks in Computational Social Science to be held at @netsciconf.bsky.social in Maastricht, June 2-6. We welcome contributed presentations on CSS research using understudied social network data (not X, Facebook etc.!): galeale.github.io/UNCSS/

Excited to share the tool @lajello.bsky.social & I built to predict social media participation in collective action! It moves beyond keywords, tracking activism stages across topics. See it in action with climate activism on Reddit 🌱 Check it out: arxiv.org/abs/2501.07368 @nerdsitu.bsky.social

Bureaucracy is so powerful that the UK Prime Minister, in a letter to the FT, doesn’t say, “We will cut bureaucracy,” but instead, “We will give you AI to break free from it.” How did we get here? www.ft.com/content/4d44...

who said naming game?