The current techniques pushes as many signals as possible to get a device fingerprint, which is then correlated with activity. It uses the pre-LLM machine learn

The current techniques pushes as many signals as possible to get a device fingerprint, which is then correlated with activity. It uses the pre-LLM machine learning methods to get such a picture. Sources of these signals are often abuses of protocols used to enhance UX across devices. Examples include: - Fonts installed (https://gist.github.com/szepeviktor/d28dfcfc889fe61763f3 https://gist.github.com/szepeviktor/d28dfcfc889fe61763f3) - Power saving telemetry (https://developer.mozilla.org/en-US/docs/Web/API/Battery_Status_API https://developer.mozilla.org/en-US/docs/Web/API/Battery_Sta...) - Screen size - https://developer.mozilla.org/en-US/docs/Web/API/Screen https://developer.mozilla.org/en-US/docs/Web/API/Screen - Deep Packet Inspection — not just IP addresses anymore - https://en.m.wikipedia.org/wiki/Deep_packet_inspection https://en.m.wikipedia.org/wiki/Deep_packet_inspection These are all tools meant to help users, but they can be used for collecting signals for targeted ads. I am sure there are more techniques that are just not publicly talked about. I have been asked by extended family members about blocking transmissions from devices so they cannot be tracked. There really is an illusion that the device actively listens on the microphone in order to target ads. I told them that, some places don’t need to transmit microphone in order to collect signals for device fingerprinting, and that is scarier.