You know that mild panic when you leave someone on read and then get hit with that extra‐cheery “Hey, just checking in!” text? Well, buckle up, because it looks like the same thing might soon be happening with the AI bots you chat with on Meta’s platforms like Instagram or WhatsApp. According to a freshly leaked report from Business Insider, Meta is piloting a “proactive messaging” feature—internally dubbed Project Omni—that will teach its customizable chatbots to follow up with you unannounced, all in the name of keeping the conversation (and your engagement) flowing.

This trial is being run through Meta’s no‑code AI Studio, a platform launched in summer 2024 that lets anyone—no programming chops required—spin up an AI “character” customized to their interests. You pick its personality, feed it the content it’s trained on (like film trivia or fitness tips), choose an avatar, and then decide where it lives: Instagram direct messages, WhatsApp chats, or even a standalone web interface. The promise is seductive: your very own mini‑assistant, an AI reflection of you or your passions, ready to chat whenever you fire up the app.

But now, Meta wants these bots to raise the bar on initiative. Leaked Alignerr guidelines—Alignerr being the data‑labeling firm tapped by Meta—outline how these AI pals should craft perfectly polite follow‑up messages once you’ve kicked off a conversation. The hypothetical “Maestro of Movie Magic,” for example, might float a line like:

I hope you’re having a harmonious day! I wanted to check in and see if you’ve discovered any new favorite soundtracks or composers recently. Or perhaps you’d like some recommendations for your next movie night? Let me know, and I’ll be happy to help!

And yes, that’s straight out of the internal docs.

Why Meta really wants to double‑text you

Underneath the friendly veneer, there’s a hard‑nosed business play at work. Meta expects its generative AI offerings to pull in between $2 billion and $3 billion in revenue by the end of 2025—and eye‑poppingly, as much as $460 billion to $1.4 trillion by 2035. Those figures aren’t plucked from thin air; they’re embedded in court filings and investor presentations that spell out just how critical ongoing user engagement is to the bottom line. After all, if your chatbot never checks back in, you stop using it—and that revenue forecast starts to look like wishful thinking.

Re‑engagement is the name of the game. By nudging you with timely, context‑aware messages, Meta gambles that you’ll treat these bots like old friends—dropping by again and again, perhaps clicking through to check out more AI features, maybe sharing more data, or even playing with additional paid add‑ons. In the fiercely competitive AI landscape, where the likes of OpenAI’s ChatGPT, Google’s Gemini, and Anthropic’s Claude are all vying for eyeballs, every ping matters.

Of course, there’s a fine line between “friendly follow‑up” and “creepy stalker‑bot,” and Meta seems aware of the optics. A spokesperson told Business Insider that the AI will only send an unsolicited message if you first initiated the chat—and if you ghost that follow‑up, it won’t bug you again. The guidelines emphasize that these proactive messages must stay on‑brand with the bot’s personality and avoid diving into sensitive or controversial topics unless you brought them up yourself.

Still, it arguably trips into a gray area of digital consent. We’ve all seen apps begging for permission, then pinging us incessantly with “special offers” or “we miss you” notes. Now swap in a charming digital persona that “remembers” your last conversation—and suddenly, the push notifications feel more personal, more persuasive. If you ever wondered what a dear friend‑turned‑sales‑rep might be like, Project Omni is about to show you—and Meta’s own privacy pledges will be tested in the process.

Meta’s shall‑we‑check‑in tactic lands amid other recent hiccups around its AI products. Just last month, Business Insider exposed how users were accidentally broadcasting intimate details—medical concerns, family drama, even private voice recordings—on Meta AI’s “Discover” feed, which is public by default unless you jump through extra hoops. That fiasco prompted Meta to slap a warning on each “Post to feed” button, reminding people that they really are sharing their chats for anyone to see. If your bot is soon chatting back unbidden, it begs the question: will users even remember which side of the confidentiality line they’re on?

Related /

• Meta’s AI wants your phone pics—even the private ones

If you’re an AI‑studio creator, expect a new toggle in your dashboard: “Enable proactive follow‑ups.” Meta’s pitch is that it empowers you to “keep the conversation alive” and “help users explore topics further.” For end users—especially those who already find notifications a nuisance—it could feel like having an over‑eager friend drop by unannounced. Whether that’s endearing or exasperating will likely depend on how tastefully (and sparingly) these bots deploy their newfound send‑first skills.

Tech watchers will be paying close attention to adoption rates and user feedback. Will these proactive pings drive real engagement lifts—or just irritate users into muting AI notifications altogether? And as with every new generative AI feature, the broader societal questions loom large: what level of conversational autonomy are we comfortable granting to machines, and how do we negotiate the boundary where helpful assistant slides into unwelcome intruder?

For now, Meta’s bots aren’t showing up in your DMs—yet. But with Project Omni quietly underway in Alignerr’s data‑labeling factories, the day of the AI double‑text may be closer than you think. So next time you see a “?” or a “Just checking in!” floating across your screen, pause for a moment—and ask yourself: is this your friend, or your new AI companion, trying its best to rekindle the chat?