Its data was already used to train AI, maybe they could have sold it to the big AI companies? I am guessing its too late for that. All they have is their previous name and user fondness.
But now with the usage fallen, maybe they could train their own coding models to compete with OpenAI, Anthropic, sort of like a hybrid of their previous forum + AI. AI solutions that work will get locked as "Accepted" by stack overflow like before via user vote.
I would suspect her phone first before looking for a physical device. An android phone is particularly vulnerable to being backdoored by cyber criminals that you can hire. Also any other computer or laptop in the house can similarly be compromised.
After that, the wifi thing is easy to check as you mentioned, but a dedicated listening device can also have a build in LTE radio with a SIM card. There is no upper limit on sophistication, even a completely passive device without external power is possible if the hostile actor has the money.[1]
it's not beyond the realm of possibility to train it. I emailed 30-40 people already with 37 screenshots including text generation and training. Not a single response. I thought for the time being this path is preferable over huggingspace/github. I'm not ruling out just open source and large .bin pretrained file...i'd need to think about it more.
The frontier of how good models are also shifts and will remain ahead of local models unless we hit some dead end limitation in the algorithms themselves. A ceiling so to speak on how good LLM can get before the law of diminishing returns starts to apply.
reply