Meta's new ad terms drop the word optional

Meta's ad terms have one sentence about automation that almost nobody has read. It covers the features that choose your audience, placements and objectives for you, and it says: “These features are optional, and you may choose to ‘opt in’ or ‘opt out’ of them, as applicable.”

On 30 October that sentence is gone. Meta has already published the new version. It describes its automation more broadly, as help “through the formatting, selection and generation of Ad Creative”. Then it says you are solely responsible for the ad, the creative, the targeting and the placements, “including for reviewing, approving or removing them, even when you use our automated features”.

There is nothing to sign. Under the terms, continuing to use Meta's ad tools after 30 October counts as accepting them.

What changes on 30 October

We read the current terms, dated 22 August 2025, against the new version line by line. Five changes matter to anyone running ads.

• Optional is gone. The current terms call automated setup optional. The new ones describe the automation and say nothing about opting out.
• Generation is named. Automation now explicitly covers the formatting, selection and generation of your creative, not only audience and placements.
• AI risk moves into the main contract. Today Meta's AI tools sit under separate Ad Creative Generative AI Terms. The new version drops that reference and writes the rule into the main terms: AI features “may make mistakes”, and you “assume any and all risks” of those features and any creative they produce.
• AI marks stay on. You may not alter or remove any watermark or authentication metadata that AI generated alongside your creative.
• Your tools are your problem. You are solely responsible for what any outside software or service does once you authorise it in your account. That covers the AI agents and automation tools now being connected to ad accounts.

Two smaller changes are worth knowing. Meta now says it does not warrant “the accuracy, completeness, or reliability of any metrics”, a line worth remembering the next time its reporting and your store disagree. And its permission to crawl your site now covers bots that pull updates on a schedule and test credentials Meta creates, with the content usable to train its products.

What is not new

Being responsible for your own ads is not new. The current terms already make you solely responsible for your ad content, targeting and placements, and Meta's AI tools already came with terms of their own. Read honestly, most of this is Meta moving existing rules into one contract.

The switches have not gone either. You can still turn off Advantage+ creative enhancements on an ad and set defaults for new ones, as we covered last month. What changes is the wording. The contract no longer says any of it is optional.

Why it lands now

The terms are catching up with the product. On 6 October Meta made video generation from your static images generally available in Advantage+ creative. In the same announcement it said its business assistant is beginning to test planning campaigns, generating creative, updating targeting and changing budgets in conversation.

Meta's line on the assistant is that “Previews, approvals, and controls let you decide what to hand off and what to own.” Under the new terms that sentence has a second half. Whatever you hand off, you still own.

The law is heading the same way

California's SB 1050, signed on 16 September 2026, takes effect on 1 January 2027. Ads that prominently feature an AI generated human performer, someone who demonstrates the product, voices the narration or reacts to it, must carry a disclosure substantially similar to “this performance features a synthetic performer”.

Two details from the law firm analyses matter to anyone using generated creative. Private lawsuits are allowed. And unlike New York's version, which has applied since 9 June 2026, California has no knowledge requirement, so not knowing a generated performer was in your ad is no defence.

Put that next to Meta's terms. If Meta's tools put a generated person or voice in your ad, you cannot strip the AI marking, you may need to add a disclosure, and the contract says the ad was yours all along. This is not legal advice. If your ads carry health, money or legal claims, show your counsel the new terms.

What to do before 30 October

• Read the new terms yourself. They are short, and the AI paragraph sits near the end.
• Open every live ad and list what Meta is allowed to generate on it: text, images, video, enhancements. Keep what you would sign. Switch off the rest.
• On anything with a price, a guarantee or a regulated claim, turn creative generation off for that ad.
• Keep a dated record of what you approved, including any generated versions you chose to keep. That record is how you show what you actually signed off.
• If an agency or a tool works inside your account, write down who approves what before anything launches. The new terms make the tool's mistakes yours.
• If any ad uses an AI generated person or voice and reaches California, plan the disclosure now so it is ready for 1 January.

The obvious objection

Here is the best case against all this. Advertisers have always been responsible for their ads, and platform contracts always protect the platform. This is Meta's lawyers tidying up, and nothing in Ads Manager changes on 30 October.

Mostly true, and that is the point. The product is now built to write, animate and rearrange your ads, and the contract has been rewritten to say every result is yours. That works when a person reads what runs before it runs. It fails in the account where nobody does.

Automation can make the ad. The terms say a person has to own it. Make sure one does.