Sama Enhances AI Security with Expanded Model Integrations

Sama has announced enhancements to its Generative AI (GenAI) solution, now integrating with leading large language models (LLMs) such as Microsoft, Google, and Amazon. This development allows Sama to securely access and retrieve data, minimizing exposure risks and easing the implementation of human-in-the-loop (HITL) components for responsible AI development.

According to Duncan Curtis, SVP of AI product and technology at Sama, these integrations ensure secure data transfer to dedicated workforces, enhancing efficiency and quality without compromising responsibility. Sama’s platform evaluates prompts and model responses across various dimensions and supports Retrieval-Augmented Generation (RAG) for improved model accuracy.

Sama plans to expand its integration capabilities further, aiming to reduce project startup times and scale to all project sizes. The HITL approach, employing over 5,000 annotators, provides crucial feedback throughout the model development process. The company's solutions aim for regulatory compliance and ethical AI practices.

All services leverage SamaHub™, a collaborative workspace, backed by SamaAssure™, guaranteeing a 98% first-batch acceptance rate. Projects are enhanced by SamaIQ™, offering insights into model vulnerabilities. Sama continues to lead in developing responsible, secure AI solutions.