Meta Ads AI Connectors Now Offer Granular Permission Controls and Account-Level Access Limits
TLDR
Meta has improved security controls for its AI Connector MCP, allowing advertisers to limit which ad accounts, pages, and business portfolios connect to Claude, ChatGPT, or Perplexity, and to granularly control which actions AI agents can perform.
Optimixed Overview
Optimixed's original report on this story.
Jon Loomer explains recent improvements to Meta's AI Connector Model Context Protocol (MCP), which links your Meta ads account to large language models. The connector now lets you selectively opt in specific Facebook pages and Instagram accounts instead of automatically exposing all current and future assets, and control which of nearly 100 tools—from campaign creation to read-only reporting—an AI agent can access. Meta has also added account-level permission controls via Business Settings, allowing advertisers to restrict sensitive actions like budget edits and campaign creation that were previously enabled by default. Loomer advises caution when connecting multiple client accounts to AI agents and emphasizes the importance of verifying AI recommendations before implementation.