Hello, Be My Eyes Community!

Be My AI (powered by OpenAI's GPT-4 image-to-text model) has proven to be an effective tool for users who are blind or have low vision. However, as part of beta testing, OpenAI’s engine may blur faces or omit facial descriptions if a person is in the picture because of legitimate privacy considerations. The resulting lack of facial description is frustrating and disappointing to blind and low vision users who hope that AI will provide the same helpful information a sighted person obtains from just looking around. We share that sentiment.

To be very clear – we aren’t talking about facial identification. Identification raises a lot of very complicated legal issues that will take more time to resolve. This is about basic facial and human descriptions.

Lack of basic facial description makes it hard for blind people to check their own appearance or enjoy social media and family photos. The insufficiency of critical detail diminishes participation in entertainment and media. Our beta testers are telling us that the omission of facial details further exacerbates the cultural gaps people in our community wish to close. We agree.

AI is a huge new tool requiring new guard rails, and right now society is in the process of inventing the right kind of protections. Be My AI runs on OpenAI’s GPT-4 model, and we both must follow an emerging patchwork of national and international laws. We’re trying to be legally diligent and responsible about privacy, regulation, and user experience – while strongly representing our community’s desire to know as much about the world around them as possible. Including faces. These are complex issues that we're committed to working on with our partners. 

And while it's possible we may not be able to fully enable facial description in the short term, we are actively pursuing long-term solutions with OpenAI. Be My Eyes also plans to work with blindness organizations and regulators to make sure the needs of the blind community are fully considered when debating, creating, and enforcing policy frameworks around AI.

In the meantime, we want you to know that as we pursue multiple solutions, we may temporarily alter, augment, or otherwise “break” our evolving AI app. Be My AI photos containing images of people occasionally may not process correctly. Be My AI might even refuse to process some images altogether. These are, of course, unintended and unexpected consequences of our continuing efforts to find a solution, but they are also a natural part of solving complicated technical and societal issues. We are working hard to limit the disruptions and to remedy them as quickly as we can. We also appreciate you letting us know when they happen, because our strong desire from day one has been to allow our community to bend this technology to their needs. 

Working with the incredible team at OpenAI, we will figure this out - ethically, responsibly, and technically. For now, we thank you for your patience and understanding, and we hope you are continuing to enjoy all the other great ways Be My AI can make the world more accessible.

If you have any questions or need help, please don't hesitate to reach out to our support team at support@bemyeyes.com! We're here to help!

Thank you!