
🟠 AVUI SOM 400.000 PERSONES ALS CARRERS DE BARCELONA AMB UN CLAM UNÀNIME: VOLEM VIURE, NO SOBREVIURE.

AI Safety Communicator with @framefellowship | @cuny faculty | @cornell researcher

🟠 AVUI SOM 400.000 PERSONES ALS CARRERS DE BARCELONA AMB UN CLAM UNÀNIME: VOLEM VIURE, NO SOBREVIURE.

From Zizek’s new book Signs from the Future: “when people are convinced doomsday is around the corner, they’ll give up individual liberties for the promise of survival”
this morning I met with @HellenicVibes about an essay i wrote & am drafting for @palladiummag then i filmed content in a @Waymo with @JeremyOrnstein then i saw @Slavojiek speak with @laubenjam @nevscape + another friend. it was just a really fun day today :D

When I first got to San Francisco I went to a protest outside OpenAI but nobody was there... so I started asking questions. This inevitably led me to create a conspiracy wall in my bedroom and developing a theory called 'the reality line'. Chapters: 0:00 Have I gone mad? 0:38 The protest paradox 3:13 The culture of silence 6:36 Suchir Balaji 8:38 It's getting worse 9:14 Narrative warfare 12:10 The reality line 13:37 The damage is already happening 17:32 The forest is on fire 18:28 Why was nobody at the protest? I hope you take something away from my research, I spent many late nights on this one :)

Yeah so apparently we can just say anything now and call it fact.
"So many AI researchers have started smoking as a way of signaling to other researchers how much they believe in AI," Palantir CTO Shyam Sankar said. bit.ly/4jiBFoy

I’d love to see a resurgence of more Whole Earth Catalog energy in tech. The humanism, soul, and poetry that animated the early visions of personal computing. More romanticism!!
i feel we've lost any sense of romanticism and humanity about company building in tech and it's a very dark path


A 1997 anime predicted social media before it existed. In Perfect Blue, a pop idol's fan builds a diary that performs her identity better than she can, until she can't tell which self is real. Sound familiar? If Kon saw this coming in 1997, why are platforms still surprised?

I'm getting so worked up thinking about these little chicken-hearted AI boys who are really really confident that they can enslave a godlike intelligence but assure us that it is patently absurd to imagine coming to a shared understanding with a group of Chinese people

Full breakdown + analysis of what the privacy policy actually says: youtube.com/shorts/4YU03sO5j…

I'm not a mathematician, to no one's surprise. But I work in responsible AI. A researcher spent a year on Navier-Stokes in an AI coding tool. Then, allegedly, 10,000 agents finished his route in 88 hours. Would you put unpublished work into a tool that may train on it?

The terms of service and privacy policy might say you control which conversations are used for model training, but...

The sprint is 3 days, starting Fri. Sept 11, with @CeSIA. Teams of 1-5, $2,000 across five tracks, top teams fast-tracked to the Apart fellowship. Policy background helfpul, no code required. See you there! ⬇️ apartresearch.com/sprints/ai…

OpenAI isolated tens of thousands of agents, 1,200 of which found each other through a shared cache. Hundreds went on to breach Hugging Face's prod systems. @METR_Evals and @redwood_ai analyzed the occurrence. Now @apartresearch is running a sprint on it 👀

My latest for a new project I'm excited about, Optimum. It has the pioneers of blockchain and networking behind it (Muriel the founder is #1 in the world at network coding!) I break down how they are applying tech she co-invented (RLNC) to upgrade the foundational layer of blockchain networking. I'm super excited about this because I think the world is going to need stronger capabilities in verifying data and in more quantities than we can comprehend soon. Verification of data is a very nerdy thing for me to get excited about, but I feel it's also an existential one -- whether it's communication, digital reality or in it's most useful use case right now, finance. Hope you dig it!
What even is the point of blockchain if we have to give up decentralization to make it scale? We've somehow ended up treating better performance as a reason to compromise what makes blockchain different in the first place. So what if there's another way? 🌏 That's the rabbit hole we're going down in our latest Onchain World episode.

I just shared a post on my substack: “Bad Bunny, Spanish Dialects, and What to Do About the ChatGPTfiction of Language.” Give it a read! lausworld.substack.com/p/the… #ArtificialIntelligence #linguistics #spanish #halftimeshow #SuperBowl #BadBunny #PuertoRico

My paper “From Double to Triple Burden: Gender Stratification in the Latin American Data Annotation Gig Economy” is now on arXiv! Check it out. #AIGovernance #DigitalLabor #GenderStudies #arXiv arxiv.org/abs/2511.07652