DeepMind Open-Sources SL2T Sign Language-to-Text Model

DeepMind has open-sourced SL2T, a sign language-to-text model developed with Deaf community input, enabling sign-based input for phone interfaces. The model weights and training pipeline are publicly available.

This shifts input modality economics. Voice and text have dominated interface design because they were the only viable mass-market input streams; SL2T makes a third, visually distinct modality viable for consumer hardware. For builders, this lowers the cost of accessibility compliance from bespoke R&D to model integration, and it creates pressure on device manufacturers to standardize front-camera gesture capture APIs as a baseline hardware feature. Expect downstream demand for fine-tuning tools that adapt SL2T to regional sign languages, which is currently the primary integration bottleneck. Operationally, teams building assistive interfaces can now prototype sign-based interactions without a dedicated computer vision team. The second-order effect: sign-to-text creates a new data flywheel for gesture recognition generally, which may accelerate other non-verbal interaction layers, from AR controls to silent device commands.