How Meta uses information for generative AI models

How Meta uses information for generative AI models and features

AI at Meta

We’re developing generative AI experiences and are excited to bring them to more people and businesses across Meta Products worldwide. That’s why we’re developing AI at Meta, our collection of generative AI features and experiences, like Meta AI and AI Creative Tools, along with the models that power them. We’re also making models available through an open platform to support researchers, developers and others in the community.

AI at Meta helps people solve complex problems, be more imaginative, and create something never seen before. From bringing real-time answers to chat, to helping people organize and plan for their next vacation, to giving them more ways to express themselves, AI at Meta helps people enhance their everyday activities, experiences and moments.

We are committed to building responsibly. We want you to know how generative AI works, and how we use your information to make this advancement possible.

What is generative AI?

Generative artificial intelligence (generative AI) enables people to quickly create content in new and exciting ways. For example, with generative AI people can easily create new content like images to share with friends and businesses can create text responses for questions from their customers.

Generative AI is not a database or static collection of information but computer models. These models are a form of artificial intelligence that are trained on billions of pieces of information from different types of data like text, images and audio. By studying this information, it can learn things like the relationship and associations between different types of content. That way the models are able to create new content when a person gives it instructions or a question. Generative AI can be used to create a variety of content like text, images and audio.

Models that generate text: Large language models (also called LLMs) are models trained on massive amounts of text to help them predict typical sequences, like those found in our everyday language. LLMs learn from the data using a combination of machine learning and guidance given by people who help train the models. They can be trained to do different language-based tasks, like completing a sentence, translating text into a different language, or responding to questions in a conversational way.
Models that generate images: These models are trained by looking at billions of images and their text captions, which are the descriptive text associated with the images. The model learns the association between these text descriptions and the images. Once it’s learned the association, it can create new images, like when a person enters a text description of the image they want generated.

Generative AI models help machines better understand the world and let people translate that understanding into creative expression.

Where does Meta get training information?

Since it takes such a large amount of data to teach effective models, a combination of sources are used for training. We use information that is publicly available online and licensed information. We also use information shared on Meta’s Products and services. This information could be things like posts or photos and their captions. We do not use the content of your private messages with friends and family to train our AIs unless you or someone in the chat chooses to share those messages with our AIs. There are more details on how we use information from Meta’s Products and services in our Privacy Policy.

When we collect public information from the internet or license data from other providers to train our models, it may include personal information. For example, if we collect a public blog post it may include the author’s name and contact information. When we do get personal information as part of this public and licensed data that we use to train our models, we don’t specifically link this data to any Meta account.

Even if you don’t use our Products and services or have an account, we may still process information about you to develop and improve AI at Meta. For example, this could happen if you appear anywhere in an image shared on our Products or services by someone who does use them or if someone mentions information about you in posts or captions that they share on our Products and services.

Privacy and generative AI

At Meta, we have a responsibility to protect people’s privacy and have teams dedicated to this work for everything we build.

We have a robust internal Privacy Review process that helps ensure we are using data at Meta responsibly for our products, including generative AI. We work to identify potential privacy risks that involve the collection, use or sharing of personal information and develop ways to reduce those risks to people’s privacy. Read more about our privacy programs and learn about our teams of experts.

To do this work ethically and responsibly, we have 5 foundational values at the core of how we think about AI. Learn more about our 5 pillars.

Privacy and security
Fairness and inclusion
Robustness and safety
Transparency and control
Governance and accountability

Our AI responsibility efforts are propelled by a cross-disciplinary team whose mission is to help ensure that AI at Meta benefits people and society. Our work on responsible AI is driven by our belief that everyone should have equitable access to information, services, and opportunities. Learn about our recent progress.

We’re investing so much in this space because we believe it provides benefits for people, creators and businesses around the world. That’s why we’re building AI at Meta and making models available through an open platform. To train effective models to unlock these advancements, a significant amount of information is needed. This is why we draw from publicly available and licensed sources, as well as information people have shared on Meta’s Products and services, including interactions with AI at Meta features. We keep training data for as long as we need it on a case-by-case basis to ensure an AI model is operating appropriately, safely and efficiently. We also may keep it to protect our or other’s interests, or comply with legal obligations.

We are committed to being transparent about the legal bases that we use for processing information. We believe use of this information is in the legitimate interests of Meta, our users, and other people. In the European region and the United Kingdom, we rely on the basis of legitimate interests to collect and process any personal information included in the publicly available and licensed sources to develop and improve AI at Meta. For other jurisdictions where applicable, we rely on an adequate legal basis to collect and process this data. You have rights related to how your information is used for AI at Meta. Learn more and submit requests here.

You may need to be logged into your account to see the appropriate forms for your region. You can learn about your other rights for information you’ve shared on Meta’s Products and services in our Privacy Policy.

We see many possibilities ahead for people, creators and businesses to use generative AI. We’re excited to see this technology come this far but know that there’s more to do in the future. Generative AI is a fast-moving space and we are committed to staying transparent as this technology evolves.