Vision Language Models (VLMs) are rapidly advancing in their capability to answer information-seeking questions. As these models are widely deployed in consumer applications, they could lead to new privacy risks due to emergent abilities to identify people in photos, geolocate images, etc. As we demonstrate, somewhat surprisingly, current open-source and proprietary VLMs are very capable image geolocators, making widespread geolocation with VLMs an immediate privacy risk, rather than merely a theoretical future concern. As a first step to address this challenge, we develop a new benchmark, GPTGeoChat, to test the ability of VLMs to moderate geolocation dialogues with users. We collect a set of 1,000 image geolocation conversations between in-house annotators and GPT-4v, which are annotated with the granularity of location information revealed at each turn. Using this new dataset, we evaluate the ability of various VLMs to moderate GPT-4v geolocation conversations by determining when too much location information has been revealed. We find that custom fine-tuned models perform on par with prompted API-based models when identifying leaked location information at the country or city level; however, fine-tuning on supervised data appears to be needed to accurately moderate finer granularities, such as the name of a restaurant or building.

It's not something to be sniffed at. Computers have cracked a problem that has stumped chemists for centuries: predicting a molecule's odour from its structure. The feat may allow perfumers and flavour specialists to create new products with much less trial and error. Unlike vision and hearing, the result of which can be predicted by analysing wavelengths of light or sound, our sense of smell has long remained inscrutable. Olfactory chemists have never been able to predict how a given molecule will smell, except in a few special cases, because so many aspects of a molecule's structure could be important in determining its odour.