IAB Content Taxonomy 2.1 → 2.2: Sensitive Topics & Risk Levels — Migration Notes

If you run ad stacks, operate a publishing platform, or build contextual classifiers, Content Taxonomy 2.2 is an important operational update to understand. Unlike the privacy-focused 2.1 release (which introduced the SCD extension), 2.2 adds a new, structured way to label and act on content that may be harmful or unsuitable — guided by the IAB Tech Lab and the Global Alliance for Responsible Media (GARM).

This article breaks down exactly what changed from 2.1 → 2.2, why those changes matter for buyers and sellers, the new sensitive-topic nodes, the brand suitability risk levels, and pragmatic migration guidance.

TL;DR — the essential change

• No existing categories were removed or re-tiered between 2.1 and 2.2.
• New top-level node: Sensitive Topics was introduced, containing 11 new topic nodes covering areas advertisers commonly treat as high-risk (e.g., terrorism, adult/explicit sexual content, hate speech).
• Brand Suitability & Risk: a complementary taxonomy subtree introduced four risk levels — Floor, High Risk, Medium Risk, and Low Risk — that act as attributes to indicate suitability for advertising.
• The two features are designed to work together: topic × risk = actionable suitability for programmatic systems.

Why 2.2 was created (and why timing mattered)

IAB Tech Lab released 2.2 in partnership with GARM to help the ecosystem standardize how it describes content that could be harmful or unsuitable for advertising. The pandemic and subsequent shifts in content moderation accelerated the need: advertisers began blocklisting pandemic-related and other critical topics, often inconsistently.

2.2 gives advertisers a consistent, machine-readable way to express what they will or won’t buy against — and gives publishers a clear framework to communicate the nature of their inventory. The result: more precise controls for advertisers and more monetization opportunities for publishers where content is safe. (IAB Tech Lab, 2024)

What was added: Sensitive Topics (the new nodes)

A new top-level category Sensitive Topics was created with the following child nodes. These are intended to be used by verification vendors, contextual classifiers, and publishers to mark content that could be unsuitable for some advertisers.

(Interactive Advertising Bureau, 2023)

Important: these nodes are additive — they do not replace SCD flags from 2.1. Use both signals where relevant: SCD flags indicate privacy-related risk from profiling, while Sensitive Topics indicates content suitability concerns for brands.

Brand Suitability & Risk: the four-level risk framework

2.2 introduced a small, but powerful subtree to express suitability levels. These are represented as separate nodes under Brand Suitability and Risk and are intended to be used as attributes alongside content topic nodes.

• Floor — Content that advertisers generally exclude by default. Think of absolute blocklist items that few brands will want adjacent to their ads.
• High Risk — Content that many advertisers avoid but where some niche or tolerant advertisers may accept placements with controls.
• Medium Risk — Content that requires consideration or limited controls; acceptable for a broader set of advertisers depending on context and creative.
• Low Risk — Content that is broadly acceptable for most advertisers, though some contextual nuance may still apply.

Relational IDs for these items (as extracted):

How topics and risk levels work together

The strength of 2.2 is that topic + risk = action. A verification vendor or classifier can label a page as both Terrorism (topic) and High Risk (suitability). Programmatic buyers can then write simple rules:

• block content where risk == Floor;
• reduce bids on High Risk topics unless the advertiser opts in;
• allow Low Risk content by default.

This model lets advertisers be nuanced — for example, an educational piece about historical terrorism might be tagged Terrorism + Medium Risk (allowable with context) instead of always being a floor-level block.

Practical impact by stakeholder

Publishers / Verification Vendors

• Labeling: Add Sensitive Topics labels to pages where applicable and emit the risk level as a separate attribute.
• Monetization: Use risk levels to help publishers identify inventory that can be safely monetized with certain advertisers (e.g., Low Risk educational content).

Advertisers / DSPs

• Controls: Implement simple, rule-based filters around bsr00x risk levels and combine them with topic selectors.
• Opt-in/Opt-out flows: Expose risk-level toggles in campaign setup so brands can be as strict or permissive as their policies allow.

Classifier & Platform Engineers

• Modeling: Train classifiers to output both topic labels and a risk score or discrete risk node so downstream systems can take action.
• Interface: Ensure bidstreams or verification payloads include both the Sensitive Topics node and the Brand Suitability and Risk node where relevant.

Legal & Compliance

• Policy alignment: Align internal brand safety policies to the four risk levels and document which levels are acceptable for which product lines or campaigns.

Key takeaways

• 2.2 is about operational brand safety. It introduces a focused set of Sensitive Topics and a four-level risk model that together make suitability decisions easier and more consistent across the ecosystem.
• No category churn. There were no removals or re-tierings — 2.2 is additive and designed to interoperate with 2.1 signals like SCD.
• Actionability: The topic × risk model lets advertisers be nuanced, publishers monetize more safely, and verification vendors standardize signals across platforms.

Final thought

Content Taxonomy 2.2 is a practical, collaboration-driven update that turns subjective brand-safety debates into machine-readable rules. For most teams, the work required is straightforward: ingest the new nodes, emit topic and risk labels, and map risk levels to your existing controls. The payoff is greater clarity and consistency in programmatic buying — and fewer surprises for brands and publishers alike.

References: