So most AI companion apps have ridiculous privacy policies. heres Replika's actual privacy policy (replika.com/legal/privacy): "In your conversations with you

So most AI companion apps have ridiculous privacy policies. heres Replika's actual privacy policy (replika.com/legal/privacy): "In your conversations with your AI companion, you may choose to provide information about your religious views, sexual orientation, political views, health, racial or ethnic origin, philosophical beliefs..." and: "...that information will be processed as part of your conversation data, including by third-party AI language model providers" So your sexual orientation, mental health stuff, political views, the late-night spiral at 2am aree all sitting on Replika's servers, plus whatever third-party LLM provider they pipe it through, plus their contractors, plus whoever does trust & safety review. and thats the company being relatively transparent about it. Character.AI's chat logs are already showing up in subpoenas. People dont talk to these apps the way they talk to ChatGPT. they tell them things they wouldnt tell their therapist. and that data is sitting in like 4 different companies infrastructure indefinitely. wWats weird is the technical excuse is gone. Apples Foundation Models + Image Playground + on-device vision mean you can ship a companion app with memory and images that runs entirely locally on a recent iPhone. "models too big for on-device" was true 2 yrs ago, not really anymore. So there's clearly a gap in the market. I built friendAI (friendai.pro) which I think is the safest AI companion out there.