WhatsApp can now call on Meta AI to summarize your personal chats. As shown in a GIF, you can access it by tapping the button to unfurl all of your unread messages in a chat. But instead of showing your messages, WhatsApp uses Meta AI to generate a bulleted summary of what you missed.
The feature is rolling out in English in the US, with plans to launch in more countries and languages later this year. It uses Meta’s Private Processing technology, which the company claims will prevent it and other third parties from snooping on your messages.
WhatsApp, which is owned by Meta, says its AI message summaries are optional, and the feature is turned off by default. You can also use WhatsApp’s “Advanced Privacy” setting to prevent users from using AI features in group chats. We still don’t know if WhatsApp’s AI message summaries will struggle with accuracy, which is something we saw with the launch of Apple’s AI-generated message and notification rundowns.
Over the past year, Meta has continued stuffing different AI features into WhatsApp, including a way to ask Meta AI questions from within a chat, as well as a feature that generates images in real-time. Some users have grown frustrated by the new Meta AI button in the bottom-right corner of the app that they can’t turn off or remove. Meta also sparked backlash with another change that brought ads to the app — something its founders said they never wanted to do.
The app’s Private Processing is supposed to conceal your interactions with its AI model by creating a “secure cloud environment,” preventing Meta or WhatsApp from seeing your summaries. Other people in the group chat won’t be able to see the message summaries, either.
What’s the point of this? If AI is summarizing your email/texts, then AI writes a response, then AI on the other end does the same… aren’t we missing the purpose of communication between two people by having AI do it for us? I want to chat with, and email my family and friends, not an AI assistant.
Ooh what’s that, is it like homomorphic encryption where they can process your data without seeing the input or output?
Meta says users can “direct AI to process their requests,” like for AI chat summaries, using Private Processing. If they do, the system won’t “retain access to user messages once the session is complete” so that a potential attacker can’t access them after the fact, according to the company.
Never mind. It’s utter bullshit. It’s a pinkie promise that they’ll take your data and totally not keep it.
Mark already told us people who trust him are “dumb fucks”
I am sure Faceberg properly isolated the contents and it is processed on the device to preserve the e2ee 🤡
