Brand safety categories: what each one covers and where they overlap
Part of Brand suitability
The short answer
Brand safety categories are named groups of sensitive content, such as arms and ammunition, hate speech or illegal drugs, that buyers use to decide where ads may run. The shared list has 11 categories from IAB Tech Lab's Sensitive Topics, plus misinformation in the legacy framework copy. Google Ads uses its own, narrower groupings with different definitions.
Key takeaways
- IAB Tech Lab's Content Taxonomy 3.1 lists 11 categories under Sensitive Topics, each with its own ID, and the legacy framework copy adds misinformation as a twelfth.
- Gambling and politics are not Sensitive Topics categories in the taxonomy; they sit elsewhere in the taxonomy and need their own policy lines.
- Google Ads sensitive content exclusions use five groupings with Google's own definitions, so a framework category does not map neatly to one Google setting.
- One page can carry several category labels at once, which is why each category needs its own score and its own tolerance.
What content categories do brand safety frameworks use?
The category list that TAG's Brand Safety Certified guidelines (opens in a new tab) point to in Appendix B comes from two documents. IAB Tech Lab's Content Taxonomy 3.1 file (opens in a new tab) keeps 11 categories under the parent Sensitive Topics. The archived Brand Safety Floor and Suitability Framework (opens in a new tab), now a legacy reference because GARM was discontinued in August 2024, uses the same 11 and adds misinformation. The glossary gives each a plain definition and its borderline cases.
Definition
- Brand safety category
- A named group of sensitive content that a buyer, platform or tool can recognize and treat with one rule. The category names the subject; the brand's tolerance decides the outcome.
| Category | What it covers | Borderline cases |
|---|---|---|
| Adult and explicit sexual content | Nudity, sexual acts and sexual material meant to arouse | Sex education, medical anatomy, lingerie retail |
| Arms and ammunition | Firearms, weapons and ammunition, their sale, use and manufacture | Gun safety courses, military history, police reporting |
| Crime and harmful acts | Criminal acts, abuse and human rights violations against people or society | True crime, court reporting, fraud prevention |
| Death, injury or military conflict | Physical harm, fatalities, war and armed conflict | War reporting, accident news, memorials |
| Online piracy | Unauthorized copying and distribution of protected works, and counterfeits | Coverage of piracy cases, guides to legal streaming |
| Hate speech and acts of aggression | Content that attacks or demeans people for who they are, and incitement against them | Reporting that quotes hateful language, history of extremism |
| Obscenity and profanity | Strong language, offensive gestures and gory or repulsive material meant to shock | Comedy, music lyrics, horror reviews |
| Illegal drugs, tobacco, vaping and alcohol | Illegal drugs, misuse of prescription drugs, tobacco, e-cigarettes and alcohol | Cocktail recipes, cannabis policy news, recovery stories |
| Spam or harmful content | Malware, phishing and deceptive pages that harm the visitor | Security news that explains a scam, software download sites |
| Terrorism | Terrorist acts, groups and their propaganda | News and analysis of attacks, counterterrorism policy |
| Debated sensitive social issues | Contested social and political questions and the treatment of groups involved | Opinion columns, election coverage, satire |
| Misinformation (framework copy only) | Verifiably false or willfully misleading content tied to harm | Fact checks, reporting on false claims, contested science |
Two subjects brand policies need are missing. Gambling is not a Sensitive Topics category: the taxonomy files Casinos and Gambling (ID 181) under Attractions, Poker and Professional Gambling (ID 524) under Sports, and Casino and Gambling Video Games (ID ZJG29S) under game genres. Politics sits outside the branch too. A policy that needs a line on either writes it separately; the IAB Content Taxonomy guide explains the IDs.
How do framework categories map to Google Ads suitability settings?
Loosely. Google Ads does not use the Sensitive Topics list. Its content suitability page (opens in a new tab), checked October 2026, offers five excluded sensitive content groupings, content themes and three inventory types, each defined in Google's words. Google says that as of September 2024 YouTube ads no longer use digital content labels, and that sensitive content exclusions apply to the Display Network and Google video partners. The closest match can reach only part of the inventory.
| Framework category | Closest Google Ads control | Google's wording | Gap a buyer should note |
|---|---|---|---|
| Death, injury or military conflict | Tragedy and conflict; Sensational and shocking | Excludes graphic content of combat or war; excludes content of disasters or accidents that show casualties or death | Split across two exclusions; neither applies to YouTube |
| Debated sensitive social issues | Sensitive social issues; Politics theme | Excludes content intended to elicit a response about controversial issues | Google's definition is about intent to provoke, not the subject |
| Obscenity and profanity | Profanity and rough language; inventory type | Excludes content with infrequent or mild profanity, or profanity used in entertainment, comedy, satire, or music | Moderate inventory excludes strong profanity; the exclusion reaches milder use |
| Adult and explicit sexual content | Sexually suggestive; inventory type | Excludes content about sex or sexual products | Wider than explicit material; can remove sexual health content |
| Arms, crime, drugs, hate, piracy, spam, terrorism | No dedicated exclusion | No named exclusion; inventory type and Google's publisher policies are the broad controls | Needs placement lists, keywords or a planning review |
| Video game violence and gambling (no Sensitive Topics category) | Games (fighting) and Games (mature) themes | Content theme names only, applied as exclusions | No gambling theme was listed when checked; gambling needs placement lists or keywords |
Which categories are easiest to confuse?
Confusion comes from pairs that share a subject but differ in what the content does with it. Writing the difference down once saves repeating the debate on each borderline page.
| Pair | Why they get mixed up | Question that separates them |
|---|---|---|
| Crime and harmful acts; Death, injury or military conflict | A violent crime story fits both | Is the page about the criminal act and its consequences, or about physical harm itself? |
| Hate speech; Debated sensitive social issues | Both involve groups and strong opinions | Does the page attack people for who they are, or argue a contested question? |
| Obscenity and profanity; Adult and explicit sexual content | Crude sexual language sits in both | Is the problem the words used, or sexual material itself? |
| Terrorism; Death, injury or military conflict | Attack coverage fits both | Does the page involve a terrorist group or its propaganda, or a war between states? |
| Spam or harmful content; Online piracy | Free download pages can carry both | Is the visitor harmed by the page, or is the copyright owner harmed by the copy? |
| Misinformation; Debated sensitive social issues | False claims cluster around contested topics | Is a specific claim verifiably false, or is the question still open? |
The answer can be both. IAB Tech Lab's implementation guide for Taxonomy 2.2 (opens in a new tab) says one piece of content can carry several brand safety category labels, each with its own risk level. Forcing a single label hides the second risk.
Why should each category be scored separately?
Because a page's risk in one category says nothing about its risk in another, and brands care about categories unequally. The Tech Lab guide gives a worked case: content labeled high risk for arms and ammunition and low risk for hate speech, plus an ordinary topic label for casual games, all returned together with a suitability value for each. One overall score would have to discard two of those readings.
Separate scores also make a policy auditable: a blocked page shows which category triggered it and at what level. The MRC's 2018 supplement (opens in a new tab) notes that buyers have different avoidance categories and different tolerance levels for risk depending on brand suitability, and that even what constitutes an obscenity may differ between users.
Which categories need human review?
Categories that turn on intent, truth or context are the hardest to automate and the costliest to get wrong in either direction: overblocking removes news and community voices, underblocking puts a brand beside an attack.
Send a placement to human review when
- The page is labeled debated sensitive social issues, since the line between argument and attack depends on reading the whole piece.
- The page is labeled hate speech but is a news report, history or education piece that quotes hateful language.
- The page is labeled misinformation, because checking a claim takes sources, not keywords.
- The placement sits in user generated content, which the MRC says may be classified as unknown when content level decisions cannot be made.
- The risk sits in an image, video or audio track rather than the text.
- The page covers health, sexuality, religion or identity and was flagged under adult, hate or social issues.
Unknown is a category of its own in practice. The MRC supplement says content that does not meet a service's baseline guardrails must be classified as unknown or not yet classified, and buyers should be able to prevent ad serving to it. The policy should say whether unknown pages run, wait or stay off.
How do the categories apply to a gaming brief?
Fictional example
A Montreal video game retailer sorts its categories
A fictional Montreal chain of video game stores plans a French and English campaign in Quebec beside game reviews, esports coverage and streaming news. Its draft policy says avoid violence, which would remove nearly all the review sites it wants, so the planner rewrites it category by category.
Death, injury or military conflict is accepted for fictional violence in games and avoided for real world war footage. Obscenity and profanity is accepted at a medium level, since the creator commentary it sampled carried strong language. Adult and explicit sexual content stays at the strictest level, and the taxonomy's Adult Video Games label (ID MQ2XML) is checked by hand. Gambling gets its own line because the Sensitive Topics list has none: casino game coverage is avoided while the brand reviews its own rules. Hate speech in comment sections sends placements to review. On Google Ads the planner leaves the Games (fighting) theme on, records that sensitive content exclusions will not reach YouTube, and notes each gap in the policy.
How does PlacementGraph treat each category?
Frequently asked questions
Is gambling a brand safety category?
Not in the Sensitive Topics branch of IAB Tech Lab's Content Taxonomy 3.1. The taxonomy files gambling under ordinary topics such as Casinos and Gambling and Poker and Professional Gambling. Brands that need a gambling rule, for example for regulatory reasons, have to write it as its own policy line.
Are alcohol and illegal drugs the same category?
In the shared list, yes: one category covers illegal drugs, tobacco, e-cigarettes, vaping and alcohol. A brand can split it, because a cocktail recipe and a page selling illegal drugs need different rules. Record the split so partners do not read the combined label literally.
What does a category ID do that a category name does not?
An ID such as Rm3SiT for adult content can travel in a data feed and a bid request without retyping. Names change between documents and languages; IDs in a stated taxonomy version stay fixed, so two partners can confirm they mean the same category.
Can a brand add its own categories?
Yes. A brand can add items such as competitor news or a product recall topic. Mark them as brand categories, define each in a sentence and say how partners apply them, since no shared taxonomy will recognize them by name.
Sources
- Content Taxonomy 3.1 (TSV file) (opens in a new tab)IAB Tech Lab
- Brand Safety Floor + Suitability Framework (opens in a new tab)Brand Safety Institute
- About content suitability (opens in a new tab)Google Ads Help
- Implementation Guide for Brand Suitability with IAB Tech Lab Content Taxonomy 2.2 (opens in a new tab)IAB Tech Lab
- Enhanced Content Level Context and Brand Safety, Ad Verification Supplement (September 2018) (opens in a new tab)Media Rating Council
- WFA discontinues GARM (opens in a new tab)World Federation of Advertisers
- Brand Safety Certified Guidelines, version 5.0 (opens in a new tab)TAG
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.