Meta Ads AI Connectors now come with additional security settings for working with Claude and other AI agents. Media buyers can now more precisely restrict what exactly AI can do within ad accounts.
We previously covered in detail Ads AI Connectors and connecting Claude to Ads Manager via MCP. We also separately wrote about Ads CLI for managing ads through the command line, and in our Hermes Agent guide we broke down how it works with the Meta Ads MCP Server and Ads CLI.
Since then, the feature set of Ads AI Connectors has expanded considerably. Back in May, 29 MCP tools were available, which we listed in our previous piece. Now, the connector settings in Claude let you manage permissions for nearly 100 MCP tools.
The ability to create campaigns, ad sets, and ads, as well as change budgets, targeting, and other settings, already existed before. The key change now is that Meta has added more granular control over these actions for individual ad accounts.
In Meta Business Suite > Settings > Integrations > Ads MCP server, you can see the ad accounts and catalogs available through MCP.
For each ad account, you can separately manage seven actions:
- budget changes;
- campaign creation;
- ad set creation;
- ad creation;
- targeting changes;
- creative changes;
- status changes.
For example, you can prevent Claude from creating campaigns, changing the budget, or editing targeting within a specific ad account.
There’s also more control at the Ads AI Connector setup stage. You can now separately select which Fan Pages and Instagram accounts the connector will have access to. That said, you still can’t select specific ad accounts directly during setup.
Additional restrictions can also be configured on Claude’s side. For MCP tools, you can allow execution without confirmation, block them entirely, or require confirmation before running.
The Ads MCP server isn’t yet available to all accounts.
