Google DeepMind has introduced a new sign-language-to-text model designed to help Deaf and hard of hearing users communicate through their phones, bringing sign language AI to consumer products for the first time, Qazinform News Agency correspondent reports.
The model, called SL2T, powers sign-to-text dictation in Gboard and Live Transcribe on Pixel 11, initially supporting American Sign Language (ASL) to English. Google said more devices and sign languages will follow.
The technology allows users to sign anywhere they would normally type. They can use it to search the web, draft messages and documents, or ask Gemini to complete tasks. In Live Transcribe, users can sign responses during conversations instead of typing them.
Unlike spoken language transcription, sign language translation requires AI to understand independent languages with their own grammars and vocabularies, as well as complex movements involving the hands, arms, torso, head and face.
SL2T was trained on more than 100,000 hours of data covering over 50 sign languages, with about a quarter of the data in ASL. Google said training across different languages, dialects and proficiency levels helps the model learn shared structures.
For privacy, the system does not send raw camera footage to the server. An on-device model tracks points on the signer’s body, and only the resulting geometric coordinates are sent for translation. The original video can then be discarded immediately. The company developed the system with members and organizations from the Deaf community, including through user studies and the AI Sign Language Advisory Committee.
SL2T is currently available in Gboard and Live Transcribe on Pixel 11 at no additional cost. Google said it plans to expand the technology to additional sign languages and develop sign language generation capabilities.
Earlier, Qazinform News Agency reported that Google launched video selfie verification for account sign-ins.