Guaranteed Safe AI Seminars

The monthly seminar series on Guaranteed Safe AI brings together researchers to discuss and advance the field of Guaranteed Safe AI. For further context, we recommend reading the Towards Guaranteed Safe AI paper.

Join: lu.ma/guaranteedsafeaiseminars

Apply to speak: fill this form

Recordings: playlist

Feedback: form

Thanks to Long-Term Future Fund for supporting the project for a 6 months duration in 2024.

Past events

December 2024: Compact Proofs of Model Performance via Mechanistic Interpretability – Louis Jaburi
November 2024: Bayesian oracles and safety bounds – Yoshua Bengio
August 2024: Constructability: Designing plain-coded AI systems – Charbel-Raphaël Ségerie & Épiphanie Gédéon
July 2024: Proving safety for narrow AI outputs – Evan Miyazono
June 2024: Gaia: Distributed planetary-scale AI safety – Rafael Kaufmann
May 2024: Provable AI safety – Steve Omohundro
April 2024: Synthesizing Gatekeepers for Safe Reinforcement Learning – Justice Sefas, and Verifying Global Properties of Neural Networks – Roman Soletskyi