Ryan J. Gallagher

With more people migrating to Bluesky (and my general quietness since Twitter fell apart), I want to reintroduce myself! I'm an applied scientist in online trust & safety. I specialize in detecting coordinated/harmful/deceptive networks, and bridging domain expertise with machine learning systems

We're shipping CoPE-B-S1, an open-weight decision model tuned for safety use cases. It's a significant step forward for our CoPE family of models in a few ways. A thread: blog.zentropi.ai/meet-cope-b-...

Social media loves talking about how everything is getting worse. But what’s something that you’ve seen stay exactly the same in your lifetime? I say rubber bands. They were fine back then and they’re fine now.

v1.136 is live! Introducing: Profile links We want independent writers, artists, developers, and creators to be able to build and fund sustainable projects on Bluesky. Help your followers know how to support you by adding buttons for Patreon, Ko-fi, your newsletter or website, and more.

Super cool post of how @fiftyplusone.news busted Wedgewood pollster for fake polls. blog.fiftyplusone.news/p/fiftypluso...

Excited to share my first PhD paper, “Hitting a Moving Target: Test-Time Adaptation for AI Text Detection under Continual Distribution Shift” (accepted to #NeurIPS2026)! We frame AI detection as a continual learning problem, robustifying models to shifting human/AI writing. (1/N)

NEW: The latest @pewresearch.org study about social media use in the US YouTube: 84% of adults use it Facebook: 70% Instagram: 51% TikTok: 37% WhatsApp: 33% LinkedIn: 31% Snapchat: 26% Reddit: 25% Nextdoor: 18% X (Twitter): 18% Truth Social: 4% www.pewresearch.org/internet/fac...

TikTok has run dozens of ads promoting the sale of counterfeit currency. One even directed users to a Telegram account offering a discount code for 10% off fake money: “TikTok10.” www.forbes.com/sites/iainma...

Seems relevant to that flood of journalists who in the last few weeks announced publicly that it was okay to return to X, that the platform had gotten better, and the worst aspects of it had been minimized. This is what you have endorsed.

Come vibe chase with us ✨🎧 open call for Platforms & Society special theme on ambient amplification, sonic social media, affective infrastructures, vibe chasing, slopification & more. Edited by Elena Pilipets & Marloes Geboers. Deadline: 30 Nov. More at: publicdatalab.org/2026/10/16/c... #AoIR2026

Twitter was one of the best things to ever happen to me and I'm never going back.

it's alive

We really need to make Meng's Big Data Paradox central to teaching, don't we? Researchers don't know this stuff bc they have never been exposed to it. But it is fundamental. (Link to paper PDF statistics.fas.harvard.edu/sites/g/file...)

It made something with much more brevity

"Scholars should carefully assess the research context and use those suggestive labels such as 'non-political' in a more thoughtful way ... The nonpolitical...is often theoretically rich and frequently intertwined with the political." www.tandfonline.com/doi/full/10....

must CVS lock each individual gummy candy away so I have to call an employee to free them? surely this can't be the way

Which tech platforms are going to participate in making this happen?

🗳 midterms.graze.social is @graze.social's most ambitious project to date. And as of today, it's live. Let's take a look at what the next 27 days will look like 🧵

Really excited to announce a new open weights safety model from Musubi: PolicyLM-1.7B: www.musubilabs.ai/blog/introdu...

Just for comparison to OpenAI’s reported 80% TPR @ FPR = 1%, for SynthID-Image, Google reports 100% TPR @ 0.1% FPR. I think it’s worth remarking that these are not remotely in the same ballpark of reliability despite both being forms of “watermark”

Exactly this. And in fact, many models *have* been created this way, and decades of prior machine learning systems *were* created this way, consensually. These things are choices! People are STILL working this way! It’s not just that a better world is possible — it already EXISTS. We can nurture it.

Just bumping into this again with people _reflexively_ using the word "grape" or a 🍇 emoji and not even realizing they're doing it. It's all part of the same system, where the men running these companies make platforms that don't even let people accurately describe the crimes that they enable.

Sad news everyone. Our beloved friend the computer has now become a hated enemy

This! Also, learn how to use a microphone. Don’t point it in the other direction from your mouth or stand four feet away. Talk directly into it. Handhold it if necessary. It’s your friend, it’s not gonna bite you.

If you are at a conference use. the. microphone. It doesn't matter how loud you think you are, it doesn't matter if you are a speaker or audience member. You should try to accommodate people who are hard of hearing (otherwise you're excluding them!)

We want to make it easier for creators to link out and attract financial support, so we're exploring a new feature that lets you add links on your profile. They show up as buttons under your bio.

Biggest culture shock on the west coast (bay area) is always how hostile everything is to just walking around What do you mean I have to cross a six lane road just to get to Stanford??

What do governments mean by “disinformation”—and what do they do about it? Today I’m launching the Global Disinformation Policy Database Explorer, a project I led at UT Austin. Explore: www.disinfo-policy.org

I hope the survivors who came forward years or decades ago know that they are the reason the Cornell case is resonating the way it is. Every single one of you helped pave a path for people to understand the harms of rape and institutional betrayal. Thank you. I know it’s hard. But thank you.

🚨 [Token][ization] Paper Alert 🚨 Tokenization is a wildly understudied area of language modeling despite it having effects across all of NLP. Over the past ~8 months, 32 (!) tokenizer researchers put together the most comprehensive survey of the field. Check it out!

Today on @indicator.media: Google is breaking its promise to label ads for unofficial government services providers. Travelers are paying the price.

I'm recruiting 1-2 PhD students to join our lab in Fall 2027, through either Computer Science or Information Science at the University of Colorado Boulder! Looking for people with interests in NLP plus [healthcare | literary studies | narratives | social media | etc.]. Join us! 🏔️☀️

The organization that governs the PLC Directory now formally exists as a registered Swiss Association and has taken the first steps towards being able to independently operate the directory. blog.plcred.org/3mwlphq42d227

I would be insufferable if I ever taught Python again

i have a new paper out in Social Media+Society about Trust & Safety specialists! 📝 i argue that they construct online harms by presenting themselves as advocates for users, creating risks, and publicizing this work. my goal was to make sense of the inherent instability of T&S work...

tawk. 🗣️

AI is a tool. Not a replacement.

🎉 New editorial: "Be Careful What You Prompt For: Generative AI in Computational Communication Research." It opens our special issue of 12 open-access papers, co-edited with @wrahool.bsky.social and @ewam.bsky.social 🧵 doi.org/10.5117/CCR2...

When we talk about the costs that LLMs impose on society we should not forget the fact that nothing works anymore because everyone is so scared of getting their shit scraped. I can't even use google scholar from my university or starlink because they send too much traffic.

the experience of COVID really drove this home for me. It wans't the apocalypse, life went on, but a million people died and there were enormous social costs, some of which continue to this day and then when it was done most people were like "let's never think about that again"