What policymakers need to know about foundation models

Foundation models are revolutionizing artificial intelligence (AI) with ability to apply knowledge across multiple scenarios and systems. Despite their vast potential, such as IBM’s MoLFormer-XL predicting molecular properties, they also pose risks like generating false content. A white paper from IBM Policy Lab recommends a risk-based, context-focused policy approach with profound transparency and differentiation between business models. Policymakers are urged to act swiftly to understand and mitigate risks while ensuring technology-neutral governance of AI.