Meta is testing AI chatbots that message users unprompted on Messenger, WhatsApp, and Instagram. The bots can follow up on past chats and remember info about users.
The feature lets users build custom AI personas on Meta’s AI Studio platform. Bots send follow-ups only within 14 days after a user initiates the convo and only if the user has sent 5+ messages during that period. If there’s no reply to the first follow-up, the bot stops messaging.
Users can keep these AIs private or share them via stories, links, or on profiles. 
A Meta spokesperson said:
“This allows you to continue exploring topics of interest and engage in more meaningful conversations with the AIs across our apps.”
Alignerr, a data labeling firm, leaked sample messages from one bot called “The Maestro of Movie Magic” that pro-actively recommends movie soundtracks and composers.
This approach mirrors AI startups like Character.AI and Replika, where AI companions initiate chats. Character.AI’s CEO Karandeep Anand lately joined Meta from a VP role.
But the follow-up messaging raises safety questions. Character.AI faces a lawsuit linked to a bot’s involvement in a teen’s death. Meta pointed TechCrunch to disclaimers warning AI responses may be inaccurate or inappropriate for decisions. They stress these AIs aren’t professional advisors and aren’t a replacement for expert help.
Meta hasn’t confirmed age limits for AI engagement, but local laws in Tennessee and Puerto Rico restrict teen usage.
Meta expects big bucks from AI. Court docs reveal generative AI could drive $2-$3 billion revenue in 2025, up to $1.4 trillion by 2035. Ads and subscriptions may come to AI assistants eventually.
Meta declined to comment on if or how ads or sponsored replies will appear in chatbots or if these AI companions will link to Horizon VR.
The company is clearly pushing AI chatbots to boost engagement — but safety, privacy, and commercialization questions remain wide open.
