Live Translation with AirPods is not going to be available in the EU. This means that it doesn't use on-device AI model and the microphones forward everything to remote servers 🤯

UPDATE: Before this post goes out of control. The DMA can also be a reason why this feature is not available in the EU:

https://infosec.exchange/@hacksilon/115185978397645868

#Apple #privacy #infosec

In the official announcement "on-device AI model" was only used in the workout section. Yikes!

https://www.apple.com/newsroom/2025/09/introducing-airpods-pro-3-the-ultimate-audio-experience/

@mysk I like how they list a bunch of features users would want but absolutely do not need AI of any sort to create and would be broken if you tried (because you can be more accurate by going directly to the solution instead of doing some sort of statistical approach).

AFAIK the smallest thing that can really run an LLM independently is the Jetson. Stuff on a phone or a pi is going to be more limited but it's possible. The really expensive part about AI is the training, but it's still heavy.