Artificial intelligence (AI) has become integral to modern life, though the rapid adoption of AI systems brings critical challenges, including reliability, compliance, security, ethics and trust. This is where Assured AI becomes essential, ensuring that AI systems, processes, frameworks and standards are robust and trustworthy – built with security in mind, and assured throughout the development and deployment lifecycle.
At Leonardo, the systems we build are often deployed in high-stake critical environments, making the need for reliable and accurate AI systems ever more important.
Leonardo’s Assured AI service seeks to ensure security and resilience of developed state-of-the-art AI products, mitigating cyber threats and aiding deployment of secure AI models in classified environments. Our experts explore the art of the possible, combining novel cyber assurance of AI for bleeding-edge use-cases with next-generation AI solutions – particularly those derived from Generative AI use-cases.
Our Assured AI services include:
- AI threat modelling & assurance – novel threat modelling and bespoke assurance strategies and research for assuring state-of-the-art models such as large language models (LLMs) for use in critical environments
- Generative AI – development of Secure-by-Design (SbD) solutions which use Generative AI black-box models such as LLMs for use in classified environments
- LLMs – SbD infrastructure to support deployment of LLMs in classified environments, development of LLMs to support a variety of use-cases (including RAG implementation and fine-tuning) and non-generative implementations of LLMs
- Classifiers – development of innovative classifiers for both text and computer vision use-cases in critical environments, utilising large-scale black box models and custom-built neural networks.
