I agree. If you really need it, run a local model.

not sure how a mailbox could be both zero-knowledge and have AI indexing it...... wait Google doesn't claim zero-knowledge

Duh! They are bad because they are trained on Gmail messages.

Honestly, at this point it’s nearly impossible to know whether they do or don’t read your emails, regardless of whether you opted in or out of AI. Time to transfer to Proton Mail I guess…
@protonprivacy

It should be opt in rather than opt out. I don't want a chatbot helping me write an email unless I push a button asking for it to help me.

Completely agree with you, unfortunately many big techs are making it opt in as the default. 🙈

The problem with AI is that it is not really that intelligent or useful beyond a few specific cases. I enjoy talking to an intelligent computer about engines, physics, and other interests, but I don't want them reading my private inbox. Sadly, big tech companies like Microsoft know that AI is not as cool as they say it is, and they know they need to force demand to make it profitable. In addition, Microsoft and Google are mainly business to business companies that offer consumer products for free to supply demand for their business to business products. For example, Microsoft makes more money from azure cloud servers and active directory than from selling Windows licenses, but windows creates demand for all the business products including azure that run on top of windows. If big tech genuinely believed in their AI products, then they wouldn't be worried about an off switch.