1047255454427887

We've designed Muse to have multiple layers of privacy and security protections. No automated system is perfect, so review Muse's activity and only grant the access you're comfortable with.

Muse runs on a cloud-based environment called a virtual machine ("VM"). This is the core environment where Muse keeps information you and things you create together. Muse also maintains backups to help protect your agent's history from data loss. Every Muse user's VM is isolated so that no one else's agent can access it.

In the future, Muse users will also have the choice to use a Muse Confidential VM: an environment that we're designing to keep people's information private, even from Meta.

Muse doesn't your conversations or the data in your virtual machine with Meta ad systems. This applies even if your Accounts Centre includes other Meta Products.

How Muse keeps logins safe

When you connect an app or service, credentials such as usernames and passwords are stored in a secure credentials store. Muse can use the Secure Credentials Store to complete an authorised action without the AI model seeing your password. Before Muse uses your credentials, the request goes through checks designed to confirm that:

Muse is connecting to the intended service

You connected or authorised the service

You gave Muse permission to perform that type of action

Protection against malicious instructions

Websites, emails, files and connected services can contain instructions intended to manipulate an AI agent. This is sometimes called a prompt injection attack. Muse uses multiple protections to help detect and resist these attempts. These include model training, automated detection systems, and limits on the information and services that Muse can access during a task. Important permission and security checks operate separately from the AI model, meaning that they don't depend only on Muse recognising that an instruction may be malicious.

Improving our AI models

You decide whether we can use your interactions with Muse to develop and improve our AI models. This setting is on when you first use Muse. You can change it at any time using the toggle in your Settings under Data controls . When the setting is off, your interactions with Muse will not be used to develop and improve AI at Meta . Changes to this setting also apply to interactions.

Allowing us to use your interactions in training helps us make smarter, more capable models. If you do allow this, we still remove certain categories of personally identifiable information, such as names, email addresses, phone numbers and social security numbers. We also disassociate the interactions from your account.

You control connected apps and services

You can choose to connect Muse to apps and services provided by Meta or third parties. We call these Connectors . When you set up a Connector, Muse gets new abilities, such as the ability to see your calendar or write an email.

We designed Muse to exchange with Connectors only the data that's needed for them to work for you. For example, connecting your email account allows Muse to your emails, but it won't download your whole inbox (unless you tell it to). Some Connectors can also information proactively with Muse. For example, calendar Connectors can update Muse automatically when there are changes in your schedule. We identify Connectors that do this before you connect them.

Permissions for important actions

Muse is designed to ask you to confirm before it takes certain important actions such as sending an email or making a purchase. When Muse requests permission, you can review the proposed action before allowing or denying it. Permission requirements may vary depending on the action, the Connector and the default permissions that you selected.

Muse is built to a set of safety standards that govern both what your Muse says and what it does. For example, these standards are designed to prevent your Muse from:

Producing harmful or illegal content

Assisting with unlawful activity, such as fraud, scams or obtaining illegal goods

Taking actions that could harm you or other people

Sexualising minors or facilitating harm to them

Deleting and forgetting Muse data

You can delete any message that you send to Muse, side chats with Muse and files in your virtual machine. If you want to start from scratch, you can reset Muse, which deletes all your Muse data.

After you've deleted something from Muse, Muse may still remember information that it learnt from what you deleted. You can also use the forget skill: simply ask Muse to forget a topic or set of interactions. Forgetting works by asking Muse to look for information a specific person, topic or matter in its memories and supporting files and remove it to the best of its ability. You can find out what Muse remembers you at any time by asking directly, or by viewing files such as your MEMORY.md .

If your Muse does something unexpected or concerning, you can report it so that our team can investigate and improve safety. To report a bug:

Click on Settings in the bottom left-hand corner and select Report an issue .

Shake your phone to open the reporting sheet, select Submit a report and complete the form, then select Submit .

You can also open the Chat tab, tap the icon in the top left, tap Settings , and select Report an issue .

Reporting helps us identify and address bugs and concerning behaviour.

Was this article helpful?