Josh Jacob

Co-Founder and (Incoming) CTO @ @SampuraResearch

Previously @GoogleDeepMind and @Waymo

[Josh Jacob @joshnotjacob]

@iarthsingh You’re absolutely right — and it’s worth being precise about the load-bearing claim here.

[Sampura Research @SampuraResearch]

Our goal is to advance the field of Scalable oversight for frontier models through improved human/AI “judges”.

Read more about our org’s mission here: https://t.co/IOTQ790guY

[William Isaac @wsisaac]

Very exciting news!

[Josh Jacob @joshnotjacob]

and PLEEEEASE come work with us full-time!

[Sampura Research @SampuraResearch]

We're backed by an initial $11M grant from Coefficient Giving ($7M for the first year, and another $4M pledged), and we're hiring founding technical staff in London: https://t.co/Gw1KGrCMAG

[Josh Jacob @joshnotjacob]

It's finally happening!!!!! So excited to announce our org and the road ahead.

[Rishub Jain @shubadubadub]

😊Today we’re launching @SampuraResearch - an AI Safety nonprofit building Human-AI Complementarity for Scalable Oversight - backed by an initial $11M grant from Coefficient Giving ($7M for the first year, and another $4M pledged). And, hiring for researchers in London. 🧵

[Josh Jacob @joshnotjacob]

Life these days.

[Josh Jacob @joshnotjacob]

@shubadubadub and I will be running the program, with @DavidDAfrica and @KozzyVoudouris continuing their awesome work advising the project.

[Josh Jacob @joshnotjacob]

⌛ One more day to apply to our SPAR fellowship for Fall 2026! (🔗⬇️)

We’re looking for folks excited to help us build a state-of-the-art benchmark for Judges (human or AI systems that evaluate whether models are correct/aligned).

[Josh Jacob @joshnotjacob]

https://t.co/pR5PhNj8RK

A big part of that work is building diverse, high-quality datasets that can rigorously test these judges. This is a crucial step in our org's mission to make scalable oversight work in practice, and we’re excited to have a new cohort on board.

[Steven Molotnikov @smolotnikov]

If AI is going to participate in our thinking, we need to build systems that keep humans in charge of the important choices.

I built Priori, an interface for human-AI collaboration, toward that goal last year.

[Josh Jacob @joshnotjacob]

@shubadubadub 🚀🚀🚀

[Rishub Jain @shubadubadub]

🫡 After 7 years, I've just left Google Deepmind to start an AI Safety nonprofit, around Scalable and Human Oversight (i.e. building stronger "judges")! 🇨🇦 And, I'll be at FAccT in Montreal this week! (1/4🧵)