Its not misleading, AI models (unlike English spelling check) has the potential to mine data on users and invade privacy all under the name of "training the model". I am very skeptical of AI models. Governments and big corp have the ability to exploit AI models and none of them a righteous for us to feel safe from a move like that.