On 12 December at #NeurIPS2026 in Atlanta, GA, our Technical Director, Lauren Greenspan, co-organises 'AI & Science: Evolution or Extinction?', a workshop on what scientific integrity requires as AI joins research. Is human science dead? Join the debate: aiscik.github.io

The other day, we launched PIRAMID: PrincInt's internal research division using statistical physics as an approach to building scientific foundations for ambitious mechanistic interpretability. Our small but mighty team has spent the last year building our infrastructure and research; today, we're sharing our results to date as well as our progress and targets for the next 6–12 months. Read about what we've been working on and where we're going: lnkd.in/eBqk8Ef7

Today, we're launching PIRAMID: an internal research division using the tools and techniques of statistical physics to build scientific foundations for ambitious mechanistic interpretability. Our central premise is that scalable alignment requires more than persuasive ad-hoc explanations of model behavior. It will require interpretability tools that develop alongside a scientific understanding of the structure of data, learning, and representations. Read more about our work: lnkd.in/eNTxH9Tv

OpenAI is looking for a diverse set of experts for help with red teaming AI models. openai.com/blog/red-teaming-…