Meta has begun expanding how information shared by outside businesses can be used, allowing data about purchases, games, and other customer activity to shape Facebook Feed content and Meta AI responses in addition to advertisements.
The policy change, announced June 9, was set to begin taking effect in July 2026 in the US and several other countries. Meta says it is not collecting a new category of information, but it is broadening how existing data can be used while removing a control that allowed users to disconnect business-shared activity from their accounts.
Enterprises that send customer information to Meta through advertising, measurement, or other integrations should review whether their disclosures accurately describe those uses. Meta has introduced privacy protections for some Meta AI conversations, but private chat modes do not address how activity from outside businesses remains connected to an account.
One data source, more uses
In its June 9 policy announcement, Meta listed purchases and games played on outside websites as examples of information businesses share with the company. Meta already uses that activity to personalize ads and will now also use it to tailor Feed content and AI responses.
The company renamed “Activity information from ad partners” as “Activity from other businesses.” The expanded setting lets users control whether the information personalizes advertising and non-advertising content.
Turning the setting off prevents Meta from using the activity for personalization. The announcement does not say that disabling it deletes the information or disconnects it from the user’s account.
Meta is also discontinuing “Your activity off Meta technologies,” which allowed users to disconnect business-shared activity from their accounts. The controls served different functions: one governed personalization, while the other governed whether the activity remained linked to an account.
Meta has not said the expanded setting will reproduce that disconnection function. It also has not specified whether the setting will be enabled by default for existing users or how people will be notified.
New AI capabilities raise the stakes
Organizations should identify which websites, applications, advertising systems, and measurement tools send customer activity to Meta. That inventory should support an enterprise AI governance program covering data access, ownership, audit records, and accountability.
Privacy and compliance teams can then compare those data flows with customer notices, consent mechanisms, contracts, and internal policies. Customers may not have been told that information sent to Meta could influence their Feed and AI responses as well as advertisements.
Meta has not published an exhaustive list of covered data categories. Its announcement also does not say the information will be used to train AI models. Personalizing a response and training the underlying model are separate data uses.
The policy carries greater weight as Meta gives its assistant more autonomy. Its push toward AI agents designed for real-world work now includes features introduced on July 24 that can connect to email and calendar apps, conduct research, create presentations, and continue scheduled tasks.
Regulators are also examining how AI assistants handle personal information. On September 11, 2025, the Federal Trade Commission opened an inquiry involving seven corporate recipients, including Meta Platforms and Instagram. The study covers safety testing, engagement monetization, and the use or sharing of information obtained through chatbot conversations. It is an industry study, not a finding that Meta violated the law.
Enterprises should watch for details about default settings, user notifications, covered data categories, and any replacement for account-level disconnection.
Read more: Reported context-aware smart glasses raise further privacy questions about information AI assistants may collect from a user’s surroundings.