On October 7th, 2025, Meta has announced that they will be using chats with Meta AI to personalize advertising and content. Since every private message and group chat on Instagram, WhatsApp, and Facebook has meta AI pushed into it, this generative AI is now able to read all of your messages, without giving the @Meta AI prompt, and adjust your content based on what you say (this includes calls and voice messages), send, read, and watch. Meta claims according to Rueters and Visual Ping, Meta has claimed it will not use topics of; religion, sexual orientation, political belief, health, racial/ethnic origins, or trade union membership to manipulate the algorithm. However, as a user of these apps, I can confirm that Meta already uses these exact topics to tailor what you see on their platforms.
What Meta says will happen: First, on December 16th, 2025, Meta AI will use instances of generative use to personalize content. However, their next statement expands on this by saying every instance of speaking to (text and voice) will also be used for “personalization” (forced advertising). Next, Meta says topics of religion, sexual orientation, political belief, health, racial/ethnic origins, and/or trade union membership will not be used for advertisements. Of course they wouldn’t add this feature, because it’s already there and has been a part of the algorithm for more than a decade. Meta also says this will be announced on October 7th, 2025, but has not made major media outlets until at the earliest of last week.
Meta has claimed that 3 things will not be coming into effect, while the last excuse contradicts the entire purpose of Meta AI and Meta’s open-source servers. First, Meta has made a statement that they will not be reading all private messages, photos, voice notes, etc. for ad-targeting, by default. This is an implication you can opt-in to be constantly watched 24/7. According to The Verge, they have said “treatment of encrypted conversations will not be changing this update.” Which is true, as only WhatsApp has encrypted messages and nowhere near on the level of Telegram. Their servers are easily accessible by the public. Meaning anyone with the programming skills can see everything you are doing without your permission, or the companies permission. At least, if you didn’t accept the Terms of Service, because you agreed to having everything you put on their platforms to be stored. The final claim is that Meta AI does not cover conversations or chats you haven’t used Meta AI in. This AI is generative, meaning it can create pictures for you based on what you ask it, and is also a language model like ChatGPT. So, this update is now also being implemented into every chat that has once used Meta AI. It will use your messages, sort through them, and then push out as much slop as possible based on what has occurred in the chat.
The fact that this very large company is collecting your lives as data and using it to push products should be extremely concerning. However, because Meta has a de facto monopoly over social media (meaning there is not many other good alternatives to Meta owned apps), people will simply accept and allow a multi-billion-dollar company to train AI using anything you say. Meta is an American company, and they are required to follow and apply the American law no matter where someone is using the app. We, collectively as people, should not allow the fourth amendment right to unreasonable search and seizures be violated in the manner that anyone can access every message we have ever sent.
Links: Meta to use AI chats to personalize content and ads from December | Reuters , Visualping , Improving Your Recommendations on Our Apps With AI at Meta
