MRC restricts property-level ad verification from brand safety claims

The advertising industry faces a significant terminology correction. On October 18, 2025, the Media Rating Council issued a policy that fundamentally redefines what qualifies as "brand safety" in digital advertising verification.

The policy addresses a persistent gap between what many verification services deliver and how they've been marketed. Property-level verification services—which analyze websites using text and keywords at the domain level—can no longer position themselves as providing "brand safety" unless they include content-level capabilities that examine images, videos, and audio.

The distinction matters more than semantics suggest. Property-level services evaluate entire domains or URLs based on text analysis and contextual classification. Content-level services analyze individual pieces of content including visual and audio elements. The difference resembles assessing a building's safety by reading the sign outside versus inspecting each room.

The policy states explicitly that "Ad Verification vendors accredited for Property-Level Ad Verification based on text or keywords as well as contextual classification based on language are not permitted to position these services as 'Brand Safety' or represent them as content-level without content-level capabilities for the specific environment." This restriction applies to service names, metric definitions, client materials, marketing releases, and all methodological disclosures.

The timeline reveals interesting context. The MRC's 2012 IAB Guidance for the Conduct of Ad Verification established methods for site context, geo-targeting, ad placement, competitive separation, and fraud detection—all using text-based analysis. That document never mentioned the word "safety." It focused on verifying campaign execution: confirming ads appeared in the correct geographic locations, at specified sizes, and without fraudulent traffic.

The 2018 Enhanced Content Level Context and Brand Safety Supplement introduced comprehensive requirements for analyzing images, videos, and audio at the URL level. This 30-page document detailed specifications for machine learning implementation, human intervention protocols, data freshness requirements, and quality control procedures. According to the October 2025 policy, "few organizations have voluntarily developed these capabilities and submitted for audit against them and even fewer have achieved accreditation."

The seven-year gap between publishing standards and enforcing terminology reflects the voluntary nature of the 2018 supplement. Vendors could choose whether to adopt content-level capabilities. Many continued offering property-level services while the "brand safety" terminology proliferated across the industry.

The new policy provides vendors a six-month grace period to comply. After April 18, 2026, the MRC "will no longer continue accreditation of any vendor that positions Property-Level Ad Verification based on text or keywords as well as contextual classification based on language as Brand Safety or content-level measurement in any way until these references are removed or altered in compliance with this policy."

Vendors can still use "suitability" language for property-level services with appropriate disclosures about measurement methodology. The policy distinguishes between brand suitability—a broader concept that property-level checking can address—and brand safety, which now requires content-level analysis.

The policy mandates five specific disclosures that property-level vendors must include in all reporting and methodological documentation. First, vendors must state explicitly "that images, video and audio are excluded from measurement and consideration with direct disclaimer that this does not provide content-level Brand Safety measurement or assurance." The disclosure must explain the limitations of their ad verification approach.

Second, vendors must clarify "that the Ad Verification Service represents Page, Site, App or Property-Level Ad Verification only as well as reporting of when unclassified URLs or subdomains are 'rolled up' to or assume domain level classifications." This addresses practices where vendors check a subset of pages then apply those classifications broadly across entire domains.

Third, they must reveal "Time to Live (TTL) policies and crawl rates including known limitations." Content changes over time. Classifications based on weeks-old data may not reflect current page conditions. Advertisers need transparency about data recency.

Fourth, vendors must disclose "Sampling and Non-Sampling Error associated to Ad Verification classifications." Automated systems make mistakes. Understanding error rates helps advertisers evaluate the reliability of classification data.

Fifth, they must acknowledge "Use of App Store details (ratings, reviews or descriptions) for classifications including known limitations." Some vendors classify mobile apps using App Store metadata rather than analyzing in-app content. This approach has inherent limitations that require disclosure.

Content-level verification requirements, as outlined in the 2018 supplement, encompass substantially more than keyword matching. Services must analyze images using computer vision, transcribe and analyze audio from videos, classify visual content frame by frame, examine user comments and interactions, assess thumbnail images, apply machine learning models trained on diverse datasets, and employ human review to validate automated findings.

The 2018 supplement provides extensive technical specifications. Machine learning models must use distinct training and evaluation datasets. Personnel performing human review need adequate training, supervision, and periodic performance assessments. Results must inform machine learning updates at least several times annually. Data freshness policies must account for the dynamic nature of content, particularly on social media platforms.

Social media and user-generated content present specific challenges that property-level verification cannot adequately address. The 2018 supplement dedicates substantial sections to these environments. Dynamic content requires fresh data and minimum baseline data guardrails. The supplement recommends that platforms either allow direct third-party measurement at a content level or provide validated first-party data to third-party measurers.

The policy takes a firm stance on future submissions: "We are not willing to consider new submissions for previously unaccredited services for Property-Level Ad Verification without firm commitments to develop Content-Level Brand Safety capabilities for images, video and audio and submit these to audit where possible." New vendors seeking accreditation must commit to developing actual content-level capabilities.

Existing vendors with content-level capabilities face expectations as well. The policy states that "All Ad Verifications organizations that have developed true Content-Level Brand Safety products or services will be asked to voluntarily submit them for audit as soon as is possible or feasible and should clearly distinguish the basis and capabilities of these products and services in disclosures and reporting."

The MRC signals potential future requirements: "We will also discuss with our Board whether there is reason or appetite to require Content-Level Brand Safety capabilities including image, video and audio, as well as compliance with the MRC Enhanced Content Level Context and Brand Safety Supplement as nonoptional for compliance with the IAB Guidance for the Conduct of Ad Verification Guidelines via a formal update."

Making content-level capabilities mandatory would represent a significant shift. The policy notes this would involve "industry and member input including consensus and public review as well as a grace period for audited services." The timeline remains "indeterminate," but the direction appears clear.

The policy document acknowledges that "MRC acknowledges there is confusion in the marketplace about the meaning and scope of the accreditations related to the two Guidelines/Standards we have referenced above, and that despite MRC's efforts to clarify otherwise, Property-Level Ad Verification without content-level measurement capabilities is often positioned in the marketplace as 'Brand Safety.'"

This acknowledgment suggests the MRC attempted to address the issue through clarification and disclosure requirements. The new policy represents a more direct enforcement approach: restricting terminology rather than relying on disclosure alone.

The policy emphasizes that "it should be noted that the IAB Ad Verification Guidelines contain no reference to 'safety' and this is a distinct service line from Content Level Brand Safety." The original 2012 guidelines established verification methods without using that terminology. The "brand safety" language emerged subsequently.

Recent developments demonstrate that content-level analysis is technically feasible. Adloox launched URL-level brand suitability segments for Google DV360 in October 2024. DoubleVerify extended brand suitability measurement to Meta Threads in October 2025, using AI-powered classification technology. These implementations indicate the industry possesses the technical capability to meet content-level requirements.

The timing coincides with increasing complexity in the digital advertising environment. AI-generated content presents new challenges for verification services. These sites often pass keyword-based checks while presenting quality concerns. User-generated content on social platforms changes rapidly, requiring continuous monitoring rather than periodic domain-level assessment.

The MRC will distinguish between three accreditation levels going forward. Property-level measurement with property-level reporting remains available for basic verification services but cannot use "brand safety" terminology. Content-level measurement with property-level reporting is permitted "with proper disclosure," allowing vendors to aggregate detailed findings into domain-level summaries if they clearly explain this approach. Only content-level measurement with content-level reporting qualifies for the brand safety designation.

Technical challenges exist in certain environments. Nested iframes—where ads serve through multiple layers of iframe tags—can limit verification visibility due to browser security features. The 2012 guidelines require vendors to report their "see-through rate" indicating how often they can verify content inside nested structures. Many advertisers may be unaware of these visibility limitations in their verification data.

Mobile applications present similar constraints. The 2018 supplement requires discrete content-level verification within mobile applications, "not based solely on descriptions of an application, application store information or traffic to commonly owned web properties." However, many apps lack API functionality that would enable content-level measurement. The supplement acknowledged this capability might be "aspirational" for many applications in 2018. The October 2025 policy suggests limited progress on this challenge.

The policy encourages platforms and publishers to facilitate independent verification: "Publishers and platforms are strongly encouraged to either allow direct third party measurement at a content level, or to actively provide validated first-party data at a granular level to enable independent Context and Brand Safety measurement, one of which will be required for content level accreditation of third-party ad verification solutions."

Advertisers evaluating verification vendors should consider specific questions: Does the service analyze images, videos, and audio in addition to text? How frequently do humans review automated classifications? What visibility exists into nested iframes? Can the service verify content in mobile apps? What measurement limitations exist? When were current classifications last updated?

The six-month grace period extends until April 18, 2026. After that date, vendors using "brand safety" terminology without content-level capabilities lose accreditation. The policy states plainly: "After the 6-month grace period, MRC will no longer continue accreditation of any vendor that positions Property-Level Ad Verification based on text or keywords as well as contextual classification based on language as Brand Safety or content-level measurement in any way until these references are removed or altered in compliance with this policy."

Vendors will likely update product names, service descriptions, and marketing materials during this transition. Advertisers should examine whether these changes reflect methodological improvements or primarily terminological adjustments. The new disclosure requirements should make these distinctions clearer.

Brand suitability violation rates show improvement in some regions—North America achieved 4.6% in 2024, down 15% year-over-year. Whether this reflects safer environments or better measurement capabilities remains an open question. Enhanced measurement tools may identify fewer false positives while detecting previously missed issues.

The policy development involved the MRC Board of Directors, suggesting institutional-level commitment to enforcement. The 2018 supplement emerged from collaboration among the MRC, the American Association of Advertising Agencies (4A's), the Association of National Advertisers (ANA), and the Interactive Advertising Bureau (IAB U.S.). Buyer-side trade organizations reviewed and approved it. The public had a formal comment period before adoption.

The industry invested substantial effort in creating comprehensive standards. The implementation gap between standard publication and enforcement reflects the challenges of voluntary compliance in a competitive market. The new policy shifts from encouraging adoption to requiring compliance for specific terminology.

The broader implication extends beyond individual vendors. The policy signals that industry self-regulation can evolve from guidance to requirements when voluntary adoption proves insufficient. The MRC's willingness to restrict accreditation for non-compliance represents a meaningful enforcement mechanism.

Advertisers can expect clearer distinctions between verification service capabilities going forward. The mandatory disclosures should provide better transparency about what different services actually measure versus what they cannot assess. This clarity should enable more informed decisions about verification investments and more realistic expectations about what different tools can deliver.

The seven-year period between establishing standards and enforcing terminology demonstrates both the complexity of industry coordination and the challenges of shifting established practices. The policy represents a course correction rather than introducing entirely new concepts—the technical requirements existed since 2018. The change involves making previously optional capabilities mandatory for using specific terminology that has market implications.

Summary

• February 14, 2012: IAB and MRC release Guidelines for Conduct of Ad Verification establishing property-level verification methods focused on campaign execution without mentioning "safety"
• September 2018: MRC publishes Enhanced Content Level Context and Brand Safety Supplement establishing detailed requirements for analyzing images, videos, and audio at URL level
• December 12, 2023: DoubleVerify expands brand safety measurement to YouTube Shorts demonstrating content-level analysis capabilities
• October 13, 2024: Adloox launches URL-level brand suitability for Google DV360 campaigns with content-level measurement
• October 24, 2024: Adloox expands brand safety verification to Facebook and Instagram using machine learning classification
• October 18, 2025: MRC issues policy restricting "brand safety" terminology to vendors with content-level capabilities including image, video, and audio analysis
• April 18, 2026: Six-month grace period expires; vendors using "brand safety" terminology without content-level capabilities lose MRC accreditation

Summary

Who: The Media Rating Council issued new requirements for ad verification vendors, restricting which services can use "brand safety" terminology based on their technical capabilities and measurement methodologies.

What: The policy prohibits vendors from using "brand safety" terminology unless they possess content-level measurement capabilities analyzing text, images, video, and audio content. Vendors offering only property-level verification must make five specific disclosures about measurement limitations: that images, video, and audio are excluded from measurement; that services represent page or domain-level verification only; time-to-live policies and crawl rates with limitations; sampling and non-sampling errors in classifications; and any reliance on app store descriptions rather than in-app content analysis. After April 18, 2026, non-compliant vendors lose accreditation.

When: The policy was issued October 18, 2025, with enforcement beginning April 18, 2026—seven years after the MRC published the Enhanced Content Level Context and Brand Safety Supplement establishing requirements for content-level verification that remained voluntary until this enforcement action.

Where: The policy applies globally to all ad verification vendors seeking or maintaining MRC accreditation across websites, mobile applications, social media platforms, and connected television environments where digital advertising appears.

Why: The policy addresses marketplace confusion where property-level verification services analyzing only keywords and domain names were positioned as providing "brand safety" despite the 2012 IAB Ad Verification Guidelines containing no reference to "safety" and content-level analysis being defined as the standard for brand safety in the 2018 supplement. The MRC acknowledges that "despite MRC's efforts to clarify otherwise, Property-Level Ad Verification without content-level measurement capabilities is often positioned in the marketplace as 'Brand Safety,'" necessitating terminology restrictions rather than relying solely on disclosure requirements.