Chandra Shekhar

British police arrested a man after Facebook reported a picture he sent privately to his wife through Messenger. Police seized both their devices. The man had downloaded an image of a Mexican actress from Google or Bing Images (he says he doesn’t remember which engine), unaware that it was AI generated. He sent it to his wife to ask if she’d like him to buy her similar clothes. According to his account, police said the image matched a hash associated with a real Mexican actress. If the image was illegal and could be identified through its hash, it shouldn’t have been available on Google or Bing Images. They should have remove it and report it to the police. Anyone in the UK can face arrest, police investigation and seizure of their devices over something they download from the internet or say in a private conversation, even when they have no idea the content could be considered illegal. Facebook Messenger is supposed to protect private conversations through end-to-end encryption. But if Meta can scan messages on people’s devices or inside the app before encryption or after decryption, that privacy protection becomes worthless against Meta’s monitoring. The same privacy and personal safety problem applies to the ChatGPT integration for Apple Messages and Meta’s Muse spyware agent. Giving these companies access to private conversations allows them to monitor and analyse sensitive information about everyone involved, including people who never consented. End-to-end encryption doesn’t protect people when they add third party services.