We aim at nothing less than a profound reshaping of society into a model where expert legal advice is no longer a privilege but as ubiquitous to everyday life as the world's collective knowledge has become through search on the internet.
We achieve this by developing the world's best Legal Large Language Model (LLM), trained by world-leading lawyers and AI experts. Our technology is engineered specifically for the legal domain, surpassing Big Tech in accuracy, safety and reliability. We do not rely on external models: we are fully independent.
Safe Sign's Founder & CEO, Sami Kardos-Nyheim, and Co-Founder & Chief Scientific Officer, Dr. Jonathan R. Schwarz, discuss their ambition to build the world's most reliable proprietary LLM for Law - fit for the accuracy demands of the legal domain.
Safe Sign Technologies is making the world's best legal advice accessible to all.
Our AI systems are built with a data-centric approach, with data drawn from proprietary data sources and carefully reviewed by our legal team, constituting the highest quality legal dataset anywhere in AI. We use new curriculum learning algorithms during training, drastically reducing the time and cost of pre-training. Subsequently, we analyze the model using active learning and adversarial methods, ensuring additional data collection is guaranteed to improve safety & performance.
Our models are trained on the foundational properties of legal reasoning, independent of jurisdiction or particular legislation. This allows us to build systems that are customized to a wide range of legal tasks, enabling us to solve problems spanning contract negotiation, legal Q&A, dispute resolution, T&C product rankings, and many more using a single model.
We emphasize safety throughout the development process with adversarial testing and red-teaming carried out by legal experts. Our models feature out-of-distribution detection, uncertainty estimation, and agent-based roundtable discussions to provide accurate and explainable responses. We specialize our system to specific jurisdictions and drastically reduce hallucinations using Retrieval Augmentation (RAG), requiring citations to up-to-date legislation with every model response.
Our team has world-leading experts in sparsity, model distillation, and quantization, ensuring our predictive models run at real-time speed and can serve thousands of requests.
"One of the best qualified founding teams I have seen." (Leading UK VC)
Safe Sign was founded in February 2022 by a world-leading team of tenured professors, AI scholars, award-winning lawyers, mathematicians, coders and influential business leaders. The cumulative expertise from Cambridge, Harvard, Oxford, DeepMind and Lenovo made Safe Sign a globally-unique start-up from Day 1.
University of Cambridge, Founder CJLPA; ex-Allen & Overy
Read More »
Research Fellow, Harvard University; Ex-Senior Research Scientist, Google DeepMind
Read More »
University of Cambridge, Machine Learning & Legal Language Model Expert
Fellow of Law, University of Cambridge; Chair of Private Law, University of Aukland
Read More »
Professor of Law, University of Cambridge, World-leading Legal AI specialist
Senior Research Associate at University of Cambridge; ex-Harvard University
Global Head of Software and Managed Services, Lenovo; ex-Manager at Sony Electronics
UK/US attorney, London School of Economics; ex-Mishcon de Reya, ex-Baker McKenzie, ex-Sony
University of Cambridge, Expert in Consumer Protection, AI, and IP Law; ex-Oxford, ex-Baker & McKenzie
University of Cambridge, University College London