Gboard Beta 17.8.3 Prepares AI Sign-to-Text Input via Phone Camera

3 articles · Updated · Android Authority · Jul 18

Version 17.8.3.939743344-beta-arm64-v8a contains an early Sign-to-Text option in Gboard, pointing to camera-based sign language input that converts gestures into written text.
Google is building the feature with a hybrid model: video stays on the device for gesture extraction, while only raw gesture data is sent to Google's cloud AI for interpretation.
App strings show the tool will guide users on camera conditions—such as warning about poor lighting—suggesting real-time visual capture is central to how it works.
The feature appears tied to Google DeepMind's sign-language AI work, but key details remain unclear, including supported sign languages, possible device limits, and whether the teardowned feature will ship publicly.

Sources

Android Authority1d ago

Gboard Update Prepares for AI-Powered Sign-to-Text Input

android.com1d ago

How to Use Circle to Search and Google Lens on Android | Android

The Eastleigh Voice News1d ago

Google's Gboard could soon understand sign language using your phone's camera | The Eastleigh Voice

4 Sources

Will Google's new feature recognize full conversational signing, or is it limited to basic fingerspelling?

After recent AI fumbles, can Google's new sign language translator be trusted not to misinterpret users?

Beyond accuracy, is signing to a phone a practical and comfortable solution for everyday communication?

Google’s Sign-to-Text for Gboard: Current Status, 11.1% Error Rate, and the Future of Accessible AI Communication

Overview

As of July 19, 2026, Google is actively developing a Sign-to-Text feature for its Gboard application to improve digital communication accessibility. This tool, still in a work-in-progress phase, was discovered through recent APK teardowns and is not yet publicly released. The feature is designed to translate sign language captured via video into written text, with Google refining both the user experience and technical requirements for optimal performance. Specific user guidance messages found in the code show ongoing efforts to address real-world usage challenges, highlighting Google’s commitment to making digital communication more inclusive.

...