Shalev

We’re announcing two new partnerships to bring innovative energy generation and storage to our data centers: 1/ 🛰️ Space Solar: Partnering with Overview Energy to beam up to 1 GW of space solar power from orbit to Earth for around the clock power production.

@iScienceLuvr Is it big enough?

Next AI conference should be held here

@polynoamial @OpenAI Awesome. Congrats on the release Noam!

I'm a manager at @OpenAI, but with GPT-5.5 I'm a more effective IC than I've ever been. I can now write CUDA kernels like a pro. I can rely on it to run my research experiments. And we know how to make it much more powerful from here.

@iScienceLuvr @SophontAI 💪💪💪💪💪

Appreciate the mention of @SophontAI's Medmarks benchmark suite in OpenAI's blog post about ChatGPT for Clinicians and HealthBench-Professional :)

The zombie’s parent is still running, so killing the zombie doesn’t help - it respawns. - Sun Tzu

@philhchen @tkkong So excited to see what you guys build! It’s gonna be epic 😎

I’ve started a new company with @tkkong! TK is a driving force behind a lot of Ramp’s success, building much of the core product, incubating the procurement platform, and leading Ramp Labs. We’re a team of IMO and Physics Olympiad gold medalists, and we’re hiring the most talent-dense team.

@iScienceLuvr LISTEN TO THIS MAN

Everyone is misunderstanding this paper+tweet. THIS PAPER DOES NOT SHOW THAT EMF RADIATION IN YOUR ENVIRONMENT IS CHANGING YOUR GENES WIRELESSLY this paper designs and implements a new ARTIFICIAL GENE SWITCH that is controllable by EMF. This is a very impressive technology!! It will hopefully be a useful tool for biology research. But! You have to do some sort of gene therapy to an organism so that the expression of a gene in that organism can be controlled with EMF radiation. Additionally, the EMF radiation used is 1,000–100,000× stronger than ambient EMF radiation. I WANT TO BE CLEAR: It does not show you can precisely control gene expression using EMF WITHOUT any modifications to an organism. "no drugs" in the original tweet is misleading. I guess technically gene therapy isn't a drug but it's still a significant intervention. Now the mechanism identified is interesting. It does rely on this endogenous protein Cyb5b that triggers a transcription factor Sp7. Now idk if this is a pathway that could ever occur naturally, this is not studied in the paper AFAICT. However Sp7 seems to be relevant in bone formation/growth/healing. So here's my speculation (which could be totally wrong which is likely!): maybe, just maybe, with very high EMF radiation you could induce bone growth/healing without modification to an organism. But biology is complex and there's probably some reason this would never work lol

chat is this real

The future of research is autonomous. This is obvious in SF now but outside of the valley I don’t think people understand that the self-improvement loop is being achieved.

@iScienceLuvr @SophontAI

I try to avoid overhyping but... I genuinely believe if we execute well, @SophontAI could help save thousands of lives by the end of this year alone. Wish us luck!

@cloneofsimo What if I fed in your whole photo album and all photo albums you’ve ever taken, and the model knew your family tree, and everything about it, and then generated a plan for creating images that touch your heart via including unique aspects of your life?

I am obviously bullish on genai and how its transforming media generation but I dont think AI models will ever touch my heart like this photo dad took

@OfficialLoganK Yesterday

AGI wen?

@tkipf Was this mostly a data problem or algorithmic problem (or both?)

It's kind of incredible how things went from "oh man, character / asset consistency is such a hard research problem" (a bit over a year ago) to our latest models just casually perfectly composing 14(!) reference assets based on whichever incantation you throw at it.

@jon_barron @Deep_Burner The journey becomes cheaper, which means you can reach many more destinations in the same amount of time!

@jon_barron @Deep_Burner Exactly

Coding got >10x more productive last month, which means that the opportunity cost of all work that isn't coding has gone up by >10x. You should rebalance your schedule accordingly: cancel meetings, reduce admin work, write fewer emails, close Overleaf. It's time to code.

@Deep_Burner @jon_barron Yes, it’s easier and cheaper to make code, which means code itself is less valuable.

@Shalev_lif @jon_barron Ok but that would lead to a glut of code, making code less valuable.

@Deep_Burner @jon_barron He’s saying that since coding is now more efficient, you can get a lot more done, so you should spend more time coding.

@jon_barron Assigning my time to coding (as it's more productive) vs non-coding (as the value of code decreases) as a SWE

@SimonLermenAI @alex_prompter Awesome work Simon

@alex_prompter I am one of the main authors, also check out my blog post: substack.com/home/post/p-18…

Holy shit… Your anonymous internet identity can now be unmasked for $1 😳 Not by the FBI. By anyone with access to Claude or ChatGPT and a few of your Reddit comments. ETH Zurich and Anthropic just dropped a paper called “Large-Scale Online Deanonymization with LLMs” and the results are the most alarming privacy research I’ve read this year. They built an automated pipeline that takes your anonymous posts, extracts identity signals, searches the web, and figures out who you are. No human investigator needed. Fully autonomous. Works on Hacker News, Reddit, LinkedIn, even redacted interview transcripts. Here’s how bad the numbers are. On Hacker News users: 67% identified correctly. When the system made a guess, it was right 90% of the time. On Reddit academics posting under pseudonyms: 52%. On scientists whose interview transcripts were explicitly redacted for privacy: 9 out of 33 still got unmasked. The pipeline works in four steps they call ESRC. Extract identity signals from your posts using LLMs. Search for candidate matches using embeddings across thousands of profiles. Reason over top candidates with models like GPT-5.2. Calibrate confidence so when it does guess, it’s almost never wrong. The classical deanonymization method from the famous Netflix Prize attack? Nearly 0% recall across every test. LLMs didn’t just improve on old techniques. They made old techniques look like toys. When they scaled to temporally split Reddit profiles, matching a user’s old posts to their newer ones across a full year gap, the pipeline hit 67% recall at 90% precision and 38% recall at 99% precision. Meaning even a year of changed interests and different conversations wasn’t enough to hide. More reasoning compute = better deanonymization. High reasoning effort doubled recall at 99% precision in some tests. As frontier models get smarter, this attack strengthens automatically. Every model upgrade is a privacy downgrade. What makes it nearly impossible to defend against: the pipeline splits into subtasks that all look benign. Summarize a profile. Compute embeddings. Rank candidates. No single API call screams “deanonymization.” The researchers themselves say they’re pessimistic that safety guardrails or rate limits can stop it. Their conclusion is blunt: “Users who post under persistent usernames should assume that adversaries can link their accounts to real identities.” And it extrapolates. Log-linear projections suggest roughly 35% recall at 90% precision even at one million candidates. Every throwaway account. Every anonymous forum post. Every “nobody will connect this to me” comment. It’s all searchable micro-data now. And the cost to run the full agent on one target is less than a cup of coffee. Practical anonymity on the internet just died. The paper killed it with math.

@TuliMathieu Sorta kinda

"but does it scale"

@iScienceLuvr Just gotta keep it running long enough

me when i watch my reward curves during RL training