Guide · AI for your ads · For the owner whose WhatsApp chats start with an ad

What your WhatsApp ad appears next to: Meta's AI content labels, the inventory filter, block lists and delivery reports

An ad that opens a WhatsApp chat does not appear on its own. On Facebook, Instagram and Threads it sits in a feed between other people's posts, or inside someone else's video. Meta's Business Help Center says the content around ads is labelled with help from its machine learning models, and it gives advertisers settings to choose how cautious to be. This guide puts those pages into plain words: how Meta's AI labels the content, what is always kept away from ads, the three inventory filter settings, block lists and exclusions, the reports that show where your ad ran, and Meta's warning that restricting where your ad shows can limit delivery and raise costs.

Published 22 October 2026

The short version

Meta calls this brand suitability. Its overview article says the controls help advertisers place ads next to organic content and publishers that suit their brand, where organic content means content posted without being promoted as an advertisement.

The labels: Meta says content uploaded to Facebook and Instagram is classified by automated and human review, relying on its machine learning models, and given a label for how sensitive it is.
The inventory filter: Your choice of how sensitive the content next to your ad may be: Expanded, Moderate or Limited. Expanded is the default.
Block lists and exclusions: Ways to keep your ad off particular Pages, profiles and apps, away from particular content, or out of live videos and certain topics.
Delivery reports: Reports that show where your ads actually appeared, so you can decide whether any of the above is needed.

How Meta's AI labels the posts around your ad

Meta's description of its brand safety methodology says content uploaded to Facebook and Instagram “goes through automated and/or human topic and content classification. This classification relies on our machine learning models”, which use the content's properties, such as audio, video and other metadata, to judge whether it meets Meta's monetisation rules.

The labels sort content into six sensitive categories, each with a severity:

The six categories: Tragedy and conflict, debated social issues, sexual and suggestive content, objectionable activity, explicit content, and strong language.
Do Not Monetize: No ads appear next to this content at all.
Severe: Only ads from advertisers on the Expanded inventory filter.
Moderate: Ads from advertisers on Expanded or Moderate.
Mild or non-sensitive: Ads on any of the three settings, including Limited.

The page says outside human raters, trained on Meta's policies, review content and “generate the content labels we use to train our machine learning models”. Content with more views is more likely to be flagged for a second look by the models or the raters. It also says plainly that a bad placement can happen, because a label changed or because of “machine learning or human rater error”. In a feed, the filter applies to the post directly before and the post directly after your ad.

Meta's separate article on community content reviews adds: “We invest in AI technology for enforcement at scale.” It says this technology helps find content for review and in some cases takes automatic action.

Our reading

You are not choosing the posts around your ad one by one; Meta's models and raters have labelled them, and your setting decides which labels your ad may sit beside. The methodology page publishes Meta's own error measurements; those are Meta's figures, and we do not repeat them. The page itself admits the labels are sometimes wrong.

What is always kept away from ads

Meta's inventory filter article says content that does not meet its Community Standards, “such as full nudity, excessive violence, terrorist acts or misinformation identified by third-party fact-checkers”, is excluded by default. Its content inventory table lists what is excluded from ads at every setting, including content that exploits controversial political or social issues for commercial purposes, content promoting illegal, prescription or recreational drugs, shocking or excessively violent content, content rated false by a third-party fact checker, and medical claims disproven by an expert organisation.

The overview article adds that content appropriate for Facebook is not necessarily appropriate to earn money from ads, and the methodology page says ads do not appear next to content labelled Do Not Monetize.

The three inventory filter settings

Meta's article on the inventory filter says it is rolling out gradually, and that if you do not see the option you should contact your Meta account representative. It covers three groups of placements: Feed ads (Facebook and Instagram feeds, Reels feed and Threads feed), in-content ads (Facebook in-stream reels and Ads on Facebook Reels), and Audience Network, Meta's ads in other companies' apps.

Expanded: Shows ads next to content that follows Meta's Content Monetization Policies, for the most reach. Meta says you are opted into Expanded by default.
Moderate: Excludes highly sensitive content. Meta says this lowers your reach and may increase costs.
Limited: Excludes additional sensitive content and live videos. Meta says this lowers reach and can increase costs.

The content inventory table shows the pattern for each of the six categories. Take strong language: several uses of strong language can appear next to an Expanded ad only; a few uses next to Expanded or Moderate; content with no strong language next to all three. Tragedy and conflict follows the same steps, from content that depicts a distressing event (Expanded only) to content that only mentions distress (all three). The table adds that it is a high-level overview and “It is possible that select content deviates in practice.”

For Feed ads, the article lists the languages the filter works in. Arabic, Urdu, Malayalam, Hindi, Tamil, Bengali and English are among them.

Our reading

For most shops, clinics and service businesses selling through WhatsApp chats, Expanded is the starting point Meta has already chosen for you. Moderate is the setting to consider if your brand would be hurt by sitting next to a post that depicts a crime, substance abuse or a distressing event. Meta's own best-practice article gives the example of a brand selling children's products that might not want its ads within more sensitive content; a business like that is the one for which Limited is worth its cost in reach.

Where to change it

In Ads Manager, Meta's article says to create an ad, click Show more options below the Placements section of the ad set, go to Inventory filter and click Edit. You can then pick Expanded, Moderate or Limited separately for Feed ads, for in-content ads and for Audience Network. If you have opted out of a placement, you may not see its option.

To set it for everything at once, Meta's article on account-level controls says people with access to your business portfolio can also apply these controls to a whole ad account in the Meta Brand Safety and Suitability Center. Meta's article on managing the Center from Meta Business Suite says Controls shows the settings for your business portfolio and ad accounts, with Inventory filters, Publisher block lists, Topic exclusions and Content type exclusions underneath. Settings made for the ad account apply automatically when anyone in it creates an ad, “even when they choose Advantage+ placements”. They cascade: an ad set can be made stricter than the account, never looser. And the article warns that delivery for current campaigns with less restrictive settings is reset to the new account setting.

Our reading

If you let Meta's AI choose where your ad shows, an account-level filter is how you keep a limit on that choice. What Advantage+ placements picks for you is in Meta's AI in your WhatsApp ads, and the places your ad can show in One picture for every spot your WhatsApp ad can show. Who can get into your business portfolio, and with what access, is covered in A Meta business portfolio for your WhatsApp.

Block lists, topics and live videos

Meta's pages describe four further controls:

Publisher block list: Keeps your ads off specific Pages, profiles and apps. The block lists article says it applies to Audience Network apps, Facebook in-stream reels, Facebook profile feed, ads on Facebook Reels and Instagram profile feed.
Content block list: Keeps your ads away from specific content in Facebook and Instagram feeds, Reels feed and Threads feed. The article says it is available through outside brand safety companies Meta works with, and that you contact one of them to create and apply a list.
Topic exclusions: Keep Facebook in-stream reels ads and ads on Facebook Reels out of content on four topics: gaming, news, politics, and religion and spirituality. Meta says this is being released gradually in specific regions only.
Content type exclusions: Keep Facebook in-stream reels ads out of live videos, and keep Facebook in-stream reels ads and ads on Facebook Reels out of videos from publishers that did not sign up to earn money from ads.

Where Meta's pages differ

The block lists article puts content block lists on Feed ads and publisher block lists on in-content and profile placements. Meta's article on Feed and in-content controls lists them the other way round: publisher block lists under Feed ads and content block lists under in-content ads. Separately, Meta's controls overview says the exclusion for publishers that did not sign up for monetisation also covers ads on Instagram Reels, while the content type exclusions article names only Facebook placements. And the topic exclusions article says live videos can be excluded for both Facebook in-stream reels and ads on Facebook Reels, while the content type exclusions article names only in-stream reels. All five pages are undated; we read them on 22 October 2026. Check what the Center shows for your own account.

Seeing where your ad actually ran

Meta's article on delivery reports says they show the publishers, and for some placements the content, your ads appeared next to or within, and that if you see anything you consider inappropriate you can adjust your inventory filters or block lists. Reports cover Facebook in-stream reels, ads on Facebook Reels, Audience Network, Facebook Feed and Facebook Reels feed. For the two feeds, they include only the top 100 posts by impressions for the last 15 days, and the article says the information is approximate and delayed by about 3 to 5 days. Downloading a report from the Center needs a business portfolio.

The methodology page says an advertiser who notices an ad next to content they feel breaks Meta's rules or their filter can report it through a form on Meta's Help Center.

Our reading

Look before you restrict. A delivery report tells you whether there is a problem to solve; a stricter setting chosen in advance costs reach whether or not there was one.

What stricter settings cost

Meta's best-practice article opens by saying you may not need these controls, because not every business has the same needs. It then says: “The ad delivery system works best with as many options for people and placements as possible.” Restricting those options limits delivery, which could make your ads more expensive and could make it harder to spend the whole campaign budget.

It suggests asking yourself:

Ads inside other people's content: Whether ads within publishers' videos and Reels are a concern for you.
Audience Network: Whether you want your ad in other companies' apps at all.
Your customers and countries: What your customers would find inappropriate, and the different policies, laws or sensitivities in the countries you advertise in.
Publishers: Whether there are publishers you do not want your ad money to reach.
Sudden events: Whether you would want to keep your ad away from content about a tragic event, or pause a campaign altogether.

Our reading

Meta's auction rewards ads that can be shown to many people in many places; each restriction narrows that. How the auction and the learning phase work is in Why Meta shows your WhatsApp ad to one person and not another. For a business advertising in the Gulf, the line about local laws and sensitivities is the one to read twice: what counts as unsuitable next to your brand may be stricter at home than Meta's default.

What these pages do not say about WhatsApp

None of the fifteen Meta pages behind this guide names WhatsApp in its text, nor ads that open a WhatsApp chat, WhatsApp Status, Stories, the WhatsApp Business app, the UAE or any Gulf country. The inventory filter article names Feed ads, in-content ads and Audience Network, and no other placements. The pages do not say how these controls apply to an ad made from inside the WhatsApp Business app.

Our reading

These controls are about the posts and videos around your ad, not the chat it opens: the conversation that follows the tap is between you and your customer. Where ads in WhatsApp Status show is covered in Ads in WhatsApp Status, and ads made in the app in Ads made from the WhatsApp Business app.

What this means for your business

BitByteCore is a verified Meta Tech Provider (since 9 September 2026). This guide repeats Meta's pages as we read them on 22 October 2026, and is not legal advice.

Our reading

Meta's AI and its human raters label what sits around your ad; your inventory filter decides which labels you accept. Start with a delivery report, not a guess. Tighten the filter only for a reason you can name, set it once at the ad account if you want it to hold for every ad, and remember Meta's own warning that restricting where your ad can show limits delivery and can make it cost more.

Questions

Does Meta use AI to decide what my ad appears next to?

Meta's description of its brand safety methodology says content uploaded to Facebook and Instagram goes through automated and/or human classification that relies on its machine learning models, using properties such as audio, video and other metadata. Outside human raters trained on Meta's policies generate the labels used to train those models. Your inventory filter setting then decides which labels your ad may appear beside.

What is the default inventory filter?

Expanded. Meta's article on the inventory filter says you are opted into Expanded by default, which shows ads next to content that follows its Content Monetization Policies for the most reach. Moderate and Limited exclude more sensitive content, and Meta says both lower reach and may increase costs.

Can my ad still appear next to something unsuitable?

Yes. Meta's pages say there is no guarantee that all content and publishers will match your standards, and its methodology page says a bad placement can happen through label changes or machine learning or human rater error. Delivery reports show where your ads appeared, and the methodology page says you can report a problem through a form on Meta's Help Center.

Does the inventory filter work for Arabic content?

Meta's inventory filter article lists the languages the filter works in for Feed ads, and Arabic is among them, along with Urdu, Malayalam, Hindi, Tamil, Bengali, English and others.

Will a stricter setting cost me?

Meta says it may. Its inventory filter article says Moderate and Limited lower reach and may increase costs, and its best-practice article says restricting people or placements limits delivery, could make ads more expensive and could make it harder to spend the whole budget.

Does this apply when Meta picks my placements?

Meta's article on account-level controls says settings made for the ad account apply automatically to new ads even when Advantage+ placements is chosen, and that an ad set can be made stricter than the account but not looser.

Who is BitByteCore?

BitByteCore is a verified Meta Tech Provider (since 9 September 2026). This guide repeats Meta's own pages, and is not legal advice.

Sources

Meta Business Help Center: About brand suitability on Meta platforms and Audience Network, read 22 October 2026.
Meta Business Help Center: Meta's brand safety description of methodology, read 22 October 2026.
Meta Business Help Center: About community content reviews, read 22 October 2026.
Meta Business Help Center: About brand suitability controls and transparency tools, read 22 October 2026.
Meta Business Help Center: About inventory filter, read 22 October 2026.
Meta Business Help Center: Content inventory table: Feed ads and In-content ads, read 22 October 2026.
Meta Business Help Center: Use inventory filter in Meta Ads Manager, read 22 October 2026.
Meta Business Help Center: Use brand suitability controls on individual ads or on your whole ad account, read 22 October 2026.
Meta Business Help Center: Manage brand suitability controls within the Meta Brand Safety and Suitability Center from Meta Business Suite, read 22 October 2026.
Meta Business Help Center: About Feed and in-content brand safety and suitability controls, read 22 October 2026.
Meta Business Help Center: About publisher and content block lists, read 22 October 2026.
Meta Business Help Center: About topic exclusions for Facebook in-stream reels and ads on Facebook Reels, read 22 October 2026.
Meta Business Help Center: About content type exclusions, read 22 October 2026.
Meta Business Help Center: About delivery reports, read 22 October 2026.
Meta Business Help Center: Best practices for brand suitability, read 22 October 2026.

This guide describes Meta's pages as we read them, and is not legal advice. More plain-language notes like this one are on the guides page.

Want customer conversations in one place you own?

We set up a business's WhatsApp number on a system its team runs, so every customer conversation sits in one place the business owns. A short call is the easiest way to see how that would work for you.

Book a demo