Google Photos & Nano Banana 2: AI images with people close to you

Less explaining, more creating: Google is bringing its Photos service and the generative AI model Nano Banana 2 closer together, allowing users to more easily generate images featuring people and objects close to them. However, the feature requires the user to allow Google to access their personal photo collection via a function called Personal Intelligence.

Nano Banana 2 is Google's image generation model for the Gemini family and was released in February 2026. When the first version of the image AI model was launched, Nano Banana was a great success: users uploaded photos en masse to create digital miniature figures based on their photos.

Google's advantage over the competition

What may alert data protectionists, according to Google's description, is intended to significantly simplify a common use case. At the same time, it is intended to tailor the Gemini app more closely to the user – instead of being a generic tool for everyone. Google has a big advantage here because many users have already uploaded the necessary data with their photos to Google anyway. The company is also working on a Gemini app for macOS that can analyze screen content to keep up with the competition. Europeans don't need to worry for now about whether they want to use the opt-in function: Personal Intelligence is currently not yet available here.

So far, users have had to formulate detailed descriptions in their prompts and manually upload reference images when they want to depict family members in AI images, for example. However, the AI is supposed to find a tailor-made answer from the personal photo collection even for prompts like “Design my dream house.”

No use for model training

Personal Intelligence is intended to recognize who is who in the photo collection based on assigned tags and person markings. In this way, templates for family members and pets could be found. If the result is still not satisfactory, the user has the option to make corrections via text input or to select alternative reference images. Those who wonder why the AI depicts a person in a certain way, for example, can also have the sources chosen by the AI displayed.

Google emphasizes that private Google Photos images will not be used directly for model training. Currently, the function is only available for subscribers of Google AI Plus, Pro, and Ultra in the USA. It is to be released for further user groups later. Personal Intelligence, introduced in January 2026, goes beyond linking with photos: it optionally also connects Gemini with Gmail, Google Calendar, Drive, YouTube, Search, Maps, and other Google apps.

(mki)