Most of what is written about AI safety is written in English, and most of the evaluations are run in English. I work on the gap: getting governance and safety ideas to researchers, lawyers and technologists across the Arab region in their own language, and testing whether the safeguards built into language models still hold in Arabic and Hebrew.

The course is the centre of that work. Its online pilot is planned with Al Akhawayn University in Morocco, with regional editions to follow. Alongside it I build evaluation tooling and design training for the safety community itself, on writing for readers outside the field.

I came to this from journalism — 50+ bylines across Al Jazeera English, the Jewish Telegraphic Agency and The Times of Israel, including interviews with Noam Chomsky and former prime ministers — and from graduate work in Arabic and Judeo-Arabic at the Hebrew University of Jerusalem.

Training: BlueDot Impact AGI Strategy and Frontier AI Governance; Harvard CS 2881 (AI Safety); Machine Learning Specialization (Andrew Ng). BA Philosophy, McGill. Based in São Paulo; open to AI-safety communication, research and evaluation work.