Nimit Kalra

@joemelko @Meta Congrats Joe! Crushing it

very belated update: i joined @meta to help train amazing models first of many releases :)

@YafahEdelman Very large knowledge updates that don't fit in-context (and it's hard/impossible to filter down to a good subset). Also I think it's not clear if/when in-context is actually as first-class as model weights.

I don't get why everyone is talking about continual learning so much. The original GPT-3 paper was all about and how models can learn in-context, and they've only gotten better since then.

@kotekjedi_ml Great work!

New paper: We deploy Claude Code in an autoresearch loop to discover novel jailbreaking algorithms – and it works. It beats 30+ existing GCG-like attacks (with AutoML hyperparameter tuning) This is a strong sign that incremental safety and security research can now be automated.

@ivanleomk LFG congrats Ivan!

Mum I made it on TBPN holy shit

coding agents that insert breakpoints for themselves

@leonardtang_ Incredible

Hello MJ1: The World's TASTIEST Judge Model Agent verification is the bottleneck to AI's progress. The field's ability to verify visual output lags far behind that of text, especially in matters of ~taste~. So we built the world's tastiest multimodal judge model, MJ1.

⚛️ Introducing CREATE, a benchmark for creative associative reasoning in LLMs. Making novel, meaningful connections is key for scientific & creative works. We objectively measure how well LLMs can do this. 🧵👇

Check out Manya's benchmark for LLM creativity! Inspired by work on creativity in graphs (@AdtRaghunathan's "roll the dice" paper), CREATE isolates testing of creative insights for discovery. Future: understand how LLMs derive insights & how they can be better creative partners!

What is a good latent space for world modeling and planning? 🤔 Inspired by the perceptual straightening hypothesis in human vision, we introduce temporal straightening to improve representation learning for latent planning. 📑: agenticlearning.ai/temporal-strai…

Proximal is hiring research scientist interns! Our research team is working on projects touching coding agents, long-horizon RL and evals. We share our work in the public through papers and open-source releases (first ones coming soon!). DM me if interested :)

we love the AI scientific discovery category i really believe llms are going to make insane future scientific discoveries last yr ppl still wondered if llms were stupid next token predictors. glad we are over that. dm me or @hunterharloff if you're building in the space

this VC firm literally mapped out the entire AI space and no one is talking about it 🤯 huge alpha here for anyone looking to get into the space or understand the current sota of tech IMO

Excited to release PostTrainBench v1.0! This benchmark evaluates the ability of frontier AI agents to post-train language models in a simplified setting. We believe this is a first step toward tracking progress in recursive self-improvement 🧵:

@Delong0_0 nice! congrats Delong!

Proud to be part of the adventure!

@zhouxy2017 @amilabs Congrats Xingyi!

Excited to share that I've joined @amilabs as a founding member. I’ll be working on world models. It is super fun building things from scratch with such a talented team. Looking forward to the journey ahead!

Unveiling our new startup Advanced Machine Intelligence (AMI Labs). We just completed our seed round: $1.03B / 890M€, one the largest seeds ever, probably the largest for a European company. We're hiring! [the background image is the Veil Nebula - a picture I took from my backyard, most appropriate for an unveiling] More details here: techcrunch.com/2026/03/09/yan…

Can we build a blind, *unlinkable inference* layer where ChatGPT/Claude/Gemini can't tell which call came from which users, like a “VPN for AI inference”? Yes! Blog post below + we built it into open source infra/chat app and served >15k prompts at Stanford so far. How it helps with AI user privacy: # The AI user privacy problem If you ask AI to analyze your ChatGPT history today, it’s surprisingly easy to infer your demographics, health, immigration status, and political beliefs. Every prompt we send accumulates into an (identity-linked) profile that the AI lab controls completely and indefinitely. At a minimum this is a goldmine for ads (as we know now). A bigger issue is the concentration of power: AI labs can easily become (or asked to become) a Cambridge Analytica, whistleblow your immigration status, or work with health insurance to adjust your premium if they so choose. This is a uniquely worse problem than search engines because your average query is now more revealing (not just keywords), interactive, and intelligence is now cheap. Despite this, most of us still want these remote models; they’re just too good and convenient! (this is aka the "privacy paradox".) # Unlinkable inference as a user privacy architecture The idea of unlinkable inference is to add privacy while preserving access to the remote models controlled by someone else. A “privacy wrapper” or “VPN for AI inference”, so to speak. Concretely, it’s a blind inference middle layer that: (1) consists of decentralized proxies that anyone can operate; (2) blindly authenticates requests (via blind signatures / RFC9474,9578) so requests are provably sandboxed from each other and from user identity; (3) relays prompts over randomly chosen proxies that don’t see or log traffic (via client-side ephemeral keys or hosting in TEEs); and (4) the provider simply sees a mixed pool of anonymous prompts from the proxies. No state, pseudonyms, or linkable metadata. If you squint, an unlinkable inference layer is essentially a vendor for per-request, anonymous, ephemeral AI access credentials (for users or agents alike). It partitions your context so that user tracking is drastically harder. Obviously, unlinkability isn’t a silver bullet: the prompt itself still goes to the remote model and can leak privacy (so don't use our chat app for a therapy session!). It aims to combat *longitudinal tracking* as a major threat to user privacy, and its statistical power increases quickly by mixing more users and requests. Unlinkability can be applied at any granularity. For an AI chat app, you can unlinkably request a fresh ephemeral key for every session so tracking is virtually impossible. # The Open Anonymity Project We started this project with the belief that intelligence should be a truly public utility. Like water and electricity, providers should be compensated by usage, not who you are or what you do with it. We think unlinkable inference is a first step towards this “intelligence neutrality”. # Try it out! It’s quite practical - Chat app “oa-chat”: chat.openanonymity.ai (<20 seconds to get going) - Blog post that should be a fun read: openanonymity.ai/blog/unlinkabl… - Project page: openanonymity.ai - GitHub: github.com/OpenAnonymity

@tejasmanohar Game changing workflow 🙌 pulling from a custom content library is a huge unlock

Content (!!) in Hightouch is happening. I'm really pumped about this new feature 🔥 The core idea is something we call "content assembly" Ask for a campaign and we’ll make it using all your existing creative across your tools like Figma, Adobe, Dropbox, etc.