Meta is expanding how much its cross-platform chatbot can tailor answers to individual users. The company says Meta AI is gaining the ability to remember details from conversations and to draw on account information from across Meta apps for more personalized recommendations.
The changes affect users in the U.S. and Canada first, and they place Meta AI closer to chatbot products that already use memory-like features to adapt responses over time.
What Meta AI Will Now Remember
Meta says users chatting with Meta AI on Facebook, Messenger, and WhatsApp for iOS and Android in the U.S. and Canada can now tell the chatbot to remember certain things about them. Examples from Meta include a user saying they love to travel or want to learn new languages.
The company also says Meta AI can pick up on “important details” from context. In practice, that means the chatbot may use something a person mentioned earlier when answering a later request.
For example, if someone previously told Meta AI that they are vegan, then later asks for breakfast ideas, the chatbot can keep that dietary preference in mind. The feature is designed to make answers feel less generic and more connected to the user’s past interactions.
Meta says there are limits. Meta AI will not remember things from group chats. Users can also delete memories at any time.
How Personal Recommendations Are Changing
The more sensitive change is not just what users directly tell the chatbot. Meta says Meta AI will also use account information from across Meta’s apps to personalize recommendations.
That information might include the home location listed on a user’s Facebook profile or recently viewed Instagram videos. The goal, according to Meta, is to make recommendations better aligned with preferences and information already shared through its services.
Personalized recommendations will go live on Facebook, Messenger, and Instagram in the U.S. and Canada to start. A Meta spokesperson told TechCrunch there will not be an option to opt out.
That detail is likely to shape how people evaluate the rollout. A memory feature that users can delete is one thing. A broader recommendation system that uses app account information without an opt-out creates a different set of expectations around control and transparency.
What Mark Zuckerberg Said
Meta CEO Mark Zuckerberg described the change in a post on Monday, framing it as a way for Meta AI to respond based on information users have already shared.
“Meta AI will start to give you answers based on what preferences and information you’ve shared,” he wrote. “For example, it’s helped me come up with creative bedtime stories for my daughters, so if I ask it for a new one, it remembers they love mermaids.”
The example highlights the basic promise of AI personalization: fewer repeated instructions and more continuity from one interaction to the next. If a chatbot remembers a recurring preference, it can produce an answer that is more useful from the first response.
But the same example also points to the broader tradeoff. The more useful a chatbot becomes through personalization, the more it depends on storing or using information about the person interacting with it.
Why This Matters For Users
Meta AI is not a standalone experiment. It is a chatbot built across major Meta products, including Facebook, Messenger, WhatsApp, and Instagram. That means changes to its personalization model can affect how people experience AI inside apps they may already use every day.
The update has two main parts:
- Conversation memory: Meta AI can remember details users provide in chats on supported apps in the U.S. and Canada.
- Cross-app personalization: Meta AI can use account information from Meta apps, such as Facebook profile location or recently viewed Instagram videos, to personalize recommendations.
Those two parts are related, but they are not identical. Conversation memory is tied to details surfaced during chats with Meta AI. Cross-app personalization reaches into information connected to other Meta app activity and account data.
For users, the practical effect could be more relevant answers. Travel preferences, language-learning interests, dietary choices, location, and video viewing patterns can all change what a recommendation should look like.
At the same time, the lack of an opt-out for personalized recommendations may be controversial. The source article notes that Meta is presenting the new data use in positive terms, but also points to the company’s broader trust problem around data, especially involving Facebook.
The Bigger Direction For Meta AI
Meta’s update shows where consumer chatbots are moving: from one-off answer engines toward systems that remember preferences and adapt over time. The source compares Meta AI’s memory feature to similar memory features for OpenAI’s ChatGPT and Google’s Gemini.
That direction can make chatbots more convenient. If users repeatedly ask for recommendations, creative ideas, or planning help, persistent context can reduce friction. A chatbot that remembers relevant preferences can skip basic setup and move more quickly to useful answers.
But personalization also raises a simple question: how much information should an AI assistant use by default? Meta’s rollout gives users deletion control over memories, but the personalized recommendation feature, as described to TechCrunch, will not include an opt-out.
For now, the rollout starts in the U.S. and Canada. The key issue will be whether users see the change as a helpful upgrade to Meta AI or as another expansion of how Meta uses information across its apps.