Meta Tests AI Ad Campaign Controls With Human Approval

Meta is testing conversational controls that would let advertisers create campaigns, generate ad assets, and change targeting or budgets inside Ads Manager. The company says previews, approvals, and controls will let advertisers decide which tasks to delegate and which to retain.

The tests move beyond an assistant that recommends changes, though advertisers cannot hand over campaign operations to an autonomous agent today.

In its October 6 Advertising Week announcement, Meta distinguishes those experiments from an expanded Ads MCP server that can send agent-prepared work into Ads Manager as a draft. Two advertising features also have broader availability: Customer Lifecycle Strategy is available to all advertisers, and generating video from existing static assets in Advantage+ creative is generally available.

Across the execution workflows, Meta is expanding what AI can prepare while saying advertiser permissions and explicit approval will continue to govern publication.

Conversational Campaign Changes Are Still Being Tested

Meta’s proposed assistant workflow would let advertisers describe campaign work in natural language instead of performing every operation manually. Testing is beginning for planning and campaign creation, generating images and copy, updating targeting, and changing budgets through conversation.

These are test capabilities. Meta uses forward-looking language for conversational campaign execution and provides no universal access date in the announcement.

Scheduled tasks and alerts are also being tested, including recurring performance monitoring, weekly summaries, and reminders to review budgets. They could make the assistant useful between conversations, but the examples do not establish that it will independently redistribute spending or publish optimizations.

A budget-review reminder draws an advertiser’s attention to a decision; a budget change alters how a campaign spends money. Meta describes both recurring assistance and conversational actions, without promising unattended campaign management.

The shift toward “agentic” capabilities is specific: the assistant is being developed to help carry out campaign work as well as explain what an advertiser might do. How much it can execute, and under which permissions, remains central to understanding the tests.

Drafts and Permissions Define the Publication Boundary

A more concrete execution workflow appears in Meta’s expanded Ads MCP server capabilities. These integrations let large advertisers and agencies work on campaigns through AI tools they already use.

According to the company, marketers can upload creative, preview ads as they would appear on Facebook and Instagram, and examine performance through trends and detailed analysis. Work prepared by agents now arrives in Ads Manager as a draft.

People and agents can work on that shared campaign without treating the agent’s output as a published advertisement. Meta says they can co-author the campaign and publish from either Ads Manager or the connected AI surface. Every action is gated by advertiser-defined permissions, with explicit approval required before anything goes live, including when publication happens through an external tool.

The controls serve distinct purposes:

• Permissions determine which actions an agent is allowed to take.
• Previews let advertisers inspect how an ad will appear before publication.
• Drafts separate prepared campaign work from live delivery.
• Approval authorizes publication rather than assuming that a request to prepare work also grants permission to launch it.

An agent might have permission to upload an asset without having permission to publish a campaign. A preview may reveal a creative issue even when the underlying action is authorized.

The announcement establishes this intended control model without providing a complete operational specification for the approval interface. Teams should not infer particular audit logs, rollback mechanisms, or account-level approval roles from the news post alone.

The expanded MCP workflows and the assistant’s conversational campaign tests are related routes toward AI-assisted execution, but Meta describes their release status separately.

Business Memory Is Deeper; Imports Remain Experimental

Meta describes deeper business memory as an update already made to its assistant. Conversations can retain context such as previous-quarter experiments, seasonal patterns, and goals an advertiser has discussed.

Importing information from other AI tools is still being tested. That capability could bring strategy, brand voice, and brand guidelines into the assistant’s context. The existing memory update does not mean every advertiser can already import a strategy document or transfer instructions from another tool.

Retained context could spare advertisers from repeating information during campaign preparation. It also raises a review question: does that information still apply? A seasonal plan or an earlier campaign objective may be useful context without being appropriate for the next launch.

Meta is also bringing creative-performance insights from Ads Creative Studio into the assistant. The studio itself is beginning a wider rollout after initial testing; it has not been declared available to every advertiser.

Social Media Today’s reporting likewise describes the expanded creative workspace and more context-aware assistant. It supports the rollout distinction, but does not independently demonstrate that the new execution features improve campaign results.

Meta reports that advertisers adopting the assistant’s recommendations typically see 5% more conversions at the same cost per conversion than those who do not. This first-party comparison is not independent evidence that the newly announced agentic tests deliver that improvement. The announcement also lacks enough methodological detail to establish adoption of recommendations as the proven cause of the difference.

Two Advertising Features Have Broader Availability Now

The clearest generally available changes concern customer acquisition and creative production. Conversational campaign execution remains in testing.

Customer Lifecycle Strategy Is Open to All Advertisers

Meta says Customer Lifecycle Strategy is now available to all advertisers. Marketers can identify a campaign as focused on acquisition and receive recommendations for excluding existing customers, instead of assembling those exclusions from scratch.

This connects campaign intent with audience recommendations. Since an acquisition campaign is meant to reach new customers, suggestions about excluding existing buyers can support that objective. They do not guarantee that every person reached will be a new customer: Meta describes exclusion recommendations, not perfect customer identification.

Excluding an entire group of existing customers with a single Audience Label remains experimental. The availability of Customer Lifecycle Strategy does not extend to that additional shortcut.

Static-to-Video Generation Is Generally Available

In Advantage+ creative, advertisers and agencies can now generate video from existing static assets, according to Meta.

Advertisers can use assets they already have to create video material. General availability establishes access, but says nothing about whether a generated video will outperform the original image or meet a brand’s requirements without review.

Ads Creative Studio also includes video concept generation and editing tools for changing hooks, audio, clip length, text overlays, captions, and ending scenes. Those tools sit within a workspace still undergoing wider rollout.

Catalog-to-video generation has a separate availability limit. Turning catalog product images into product videos is restricted to eligible beta advertisers through Advantage+ creative in Ads Manager and the product video creator in Commerce Manager. It is a different release from broadly available static-to-video generation.

Audience Discovery Is a Promise, Not a Current Release

Meta’s AI-enabled audience discovery is planned for Detailed Targeting by the end of 2026. The proposed workflow would let advertisers describe their ideal customer in plain language, receive audience suggestions, and choose which ones to retain.

Customer Lifecycle Strategy, already available, lets advertisers express an acquisition objective and receive exclusion recommendations. Audience discovery is a forthcoming way to find targeting options.

Meta also plans to scale product recommendations that analyze an ad’s image, text, and landing page to suggest related products. Those plans are separate from the creative features explicitly described as generally available.

Advertisers’ immediate options are narrower than the announcement’s full scope. They can evaluate the two broadly available features, check whether their accounts have access to the expanding creative workspace, and assess MCP-based workflows where supported. Conversational execution and scheduled assistance remain tests.

The consequential part of Meta’s direction is the handoff between an agent’s prepared work and an advertiser’s authorization. Drafts and approvals can make that handoff explicit. Their practical value will depend on whether reviewers can clearly see what the agent proposes to change, especially when it affects targeting or spending. Meta has outlined the boundary; the next evidence to watch is how reliably its tools preserve it in actual campaign workflows.