We want to work with you
We are not hiring for full-time positions right now.
We are always open to research collaborations. If you work on alignment, interpretability, capability evaluation, scalable oversight, or adversarial robustness — and you have something concrete to work on — we want to hear from you.
We are also interested in partnerships with labs and universities working on technical AI safety. If there is a specific research question or dataset you think we should be working on together, email us.
Reach out at safeagiforum@gmail.com
We want to pay it forward
Technical AI safety is a young field. The infrastructure for training researchers, sharing results, and building a shared research culture is still being built. We want to contribute to that.
We run reading groups, paper walkthroughs, and working sessions on mechanistic interpretability, alignment, and evaluation. These are open to anyone working seriously on technical AI safety — no affiliation required.
If you are a student or early-career researcher trying to get into the field, we can help. We offer informal mentorship, access to our research network, and occasional visiting positions for people working on problems we care about.
Current Reading Groups
- Mechanistic Interpretability — Weekly
- Alignment Forum Analysis — Fortnightly
- Capability Evaluation — Monthly
To join or get involved: safeagiforum@gmail.com
