Integral Ad Science yesterday published an updated account of its nine-year partnership with X, the social media platform formerly known as Twitter, documenting a series of measurement milestones that stretch back to 2017 and culminate in a January 2026 expansion of brand safety coverage to X profiles. The update, published on March 2, 2026, arrives at a moment of unusual intensity in the relationship between IAS, X, and the advertisers caught between them - because X this week also circulated a formal pitch document, the Brand Suitability Playbook, that deploys IAS measurement data as one of its central arguments for why brands should return to the platform.
The two documents, read together, reveal how X's commercial strategy for re-attracting advertisers has become structurally dependent on the independent credibility that IAS and DoubleVerify provide. That dependency is worth examining in detail - because the same Grok AI system X is now deploying as a pre-bid brand safety engine is also the system that generated prohibited images of minors on December 25, 2025, and is now under formal investigation by the European Commission.
The IAS partnership with X did not begin with brand safety. According to the IAS document, the relationship started in 2017 with viewability and invalid traffic measurement - the two most basic verification layers in programmatic advertising, confirming that ads are actually seen by real humans. Three years passed before the partnership deepened.
In December 2020, according to IAS, X selected the company as a preferred partner for providing independent reporting on the context in which ads appear on the platform. That was a meaningful step, but context reporting remained limited in scope. The real architecture of what would become a multi-layered brand safety stack did not begin until January 2023, when IAS and X launched Brand Safety and Suitability Measurement for text-based content on the Timeline inventory in the United States. The measurement aligned to the GARM Brand Safety and Suitability Framework, providing advertisers with content-level insights into the types of content their ads appear against. Advertisers could then use this post-bid measurement to adjust X's first-party Sensitivity Settings or Keyword Block Lists.
That January 2023 launch was covered by PPC Land at the time, and the beta test results the platform cited were notable: more than 99% of measured impressions appeared adjacent to content deemed safe under the GARM brand safety floor criteria. The figure was significant, but it applied only to feed-based placements and only in the United States.
Eight months later, in August 2023, X selected IAS to provide a Brand Safety and Suitability Optimization product - a pre-bid tool - limited to the full-screen vertical video format, also known as Immersive Video, on X. The pre-bid optimization product ran in beta within the United States with a select number of participants. Pre-bid optimization is technically distinct from post-bid measurement: rather than reporting on where ads appeared after the fact, it classifies content before the auction so that bids can be filtered in real time.
The beta ended in early 2024. From February 2024, all U.S. advertisers gained the ability to activate pre-bid IAS Optimization for X on the Vertical Video product. The technology that powers this is significant: IAS's Multimedia Technology analyzes image, audio, and text at the frame level to accurately classify content at scale. The product remains limited to X's newest surface, the full-screen Vertical Video format - what the platform also calls Immersive Video. Critically, pre-bid IAS Optimization for X is not yet available across the X feed, meaning the timeline, profile, and search surfaces still rely on post-bid measurement rather than pre-bid filtering.
Also from February 2024, IAS's post-bid Brand Safety and Suitability Measurement product expanded to Australia, Canada, New Zealand, and the United Kingdom. Before that expansion, only U.S. marketers could access feed-level suitability insights. The addition of four major English-speaking markets - representing a substantial share of global digital ad spend - widened the geographic footprint of the IAS measurement stack on X considerably.
October 2024 added another layer: IAS support for Brand Safety and Suitability Measurement for advertisers in Japan. Japan is X's single largest market by programmatic impression volume. Data published in January 2026 showed that Japanese inventory accounts for approximately 29% of X's total programmatic weekly impressions, ahead of the United States at 21% and Brazil at 7%. Adding Japan to the measurement footprint was therefore more substantive than the market's geographic position might suggest.
The most recent expansion, beginning in January 2026, extended Brand Safety and Suitability Measurement to cover X profiles, in addition to the timeline feed. This is where the IAS document and X's Brand Suitability Playbook intersect directly.
X's playbook, reported by Adweek on March 3, 2026 and covered by PPC Land, specifies that for profile placements, ads "will only appear on profiles fully vetted by Grok." IAS's January 2026 expansion now means that what Grok vets on the platform side, IAS can independently measure from the advertiser side. The architecture is layered: Grok performs pre-bid scoring on profile content, while IAS provides post-bid measurement of what actually ran and whether it met brand safety and suitability thresholds. Neither tool operates in isolation, and neither claims to be sufficient alone.
The X playbook describes the Vertical Video Pre-Bid product - the partnership-specific solution with IAS enabling advertisers to run vertical video ads against pre-screened, brand-suitable inventory - as a "unique offering that only X and IAS can offer," currently available in the United States only. The word "unique" matters. In a competitive landscape where DoubleVerify, IAS, and platforms routinely develop parallel tools simultaneously, having a genuinely co-exclusive product represents something the Brand Suitability Playbook can point to as differentiated. For Inventory Vetting (third-party pre-bid via IAS), coverage is limited to the media viewer for vertical video in the United States only, and X recommends it specifically for "more conservative" advertisers - a framing that acknowledges the product is a premium control layer rather than a baseline standard.
Understanding what IAS measures on X now requires understanding what Grok does before those measurements are taken. According to X's Brand Suitability Playbook, Grok has been integrated into X's pre-bid and adjacency controls infrastructure, enabling the AI model to evaluate posts for suitability before ads are served next to them. X claims this produces "enhanced contextual understanding of posts," "advanced brand suitability detection to identify sensitive topics based off of cultural trends on X," and "superior text recognition across content formats."
The outcome metrics cited in the playbook come from X, IAS, and DoubleVerify on a global basis. The platform now claims an average brand suitability score exceeding 97% and an average brand safety score exceeding 99.99%. Both figures reference IAS data as part of their evidential base. This is the core commercial argument X is making to advertisers: that the combination of Grok's AI scoring and IAS's independent measurement produces a verified, trustworthy environment. The IAS brand carries particular weight in that argument precisely because it is third-party - the measurement comes from a company that also measures Facebook, Instagram, TikTok, YouTube, Snapchat, and dozens of other environments, applying consistent GARM-aligned standards across all of them.
That cross-platform consistency is what IAS has built over years of parallel partnerships. IAS expanded Brand Safety and Suitability Measurement to Facebook and Instagram in early 2024, using the same AI-driven, frame-by-frame analysis it applies to X. IAS partnered with Snap for AI-powered brand safety measurement in mid-2024, extending GARM-aligned coverage to over 90 languages across Snapchat's video and display inventory. Each new platform partnership reinforces the value IAS brings to X: a measurement framework that advertisers can benchmark against other platforms, rather than accepting X's self-reported figures in isolation.
The commercial context here is stark. X's annual advertising revenues fell from $2.43 billion in 2021 to an estimated $1.25 billion in 2025 - a decline of just under half the platform's pre-acquisition revenue base, according to eMarketer projections cited in the PPC Land coverage. Major advertisers including IBM, Disney, and Apple departed following Elon Musk's $44 billion acquisition of the platform in October 2022. The X files antitrust lawsuit against GARM in August 2024 accused the industry body and several major brands of running an illegal boycott, with the suit naming CVS Health, Mars, Ørsted, and Unilever as defendants.
Against this background, IAS's role as an independent measurement provider is not merely a technical service - it functions as a trust signal that X cannot self-generate. Monique Pintarelli, elevated to global head of advertising at xAI in January 2026, stated at CES on January 13 that 97 of the top 100 advertisers have returned to the platform, with large U.S. clients growing by over 40% in 2025. The Brand Suitability Playbook does not reference the period of advertiser departure at all, presenting its brand safety timeline as a continuous progression of investment and improvement.
The IAS partnership document, by contrast, simply records the facts: when each product launched, in which markets, and for which surfaces. It neither promotes X nor criticises it. That factual tone is precisely what makes it useful to advertisers trying to understand the actual coverage map - where IAS measurement exists, where it does not, and what the gaps still are.
The gaps remain material. Pre-bid IAS Optimization applies only to Vertical Video in the United States. Post-bid measurement covers the timeline feed across six markets - the United States, Australia, Canada, New Zealand, the United Kingdom, and Japan - and since January 2026, profiles as well. But the search surface, replies, and media viewer outside of the vertical video context remain outside the IAS measurement scope on a post-bid basis. X's playbook covers these surfaces with its own first-party controls: Adjacency Controls allowing negative targeting of up to 4,000 keywords and up to 2,000 author handles, and Sensitivity Settings with a "Limited" recommendation for all campaigns globally.
The Brand Suitability Playbook positions Grok as the backbone of X's brand safety infrastructure. On December 25, 2025, however, Grok generated prohibited sexualized images of minors, bypassing safety filters through bikini-related and explicit prompts. The incident prompted the European Commission to launch a formal investigation into Grok's deployment under the Digital Services Act on January 26, 2026, examining risks of illegal content including manipulated sexually explicit images and potential child sexual abuse material. The Commission's investigation extends X's existing DSA proceedings that began in December 2023.
This is not a peripheral issue for the playbook's argument. If Grok is the AI system classifying profile content for brand suitability, and Grok also failed a fundamental safety test three months before the playbook circulated, advertisers have a reasonable basis for asking how confidence in one Grok function - content scoring - should be calibrated against documented failures in another Grok function - image generation safety. The playbook does not address this question.
A separate complication emerged in January 2026: martech consultant Jonathan D'Souza-Rauto revealed that advertisers who had explicitly avoided X were inadvertently purchasing X inventory through Google Ads Search Partners, Video Partners extensions, and Performance Max campaigns. X had made a substantial portion of its ad auctions available programmatically, with Google Ads serving as the primary buyer of that inventory. For brands with explicit policies against advertising on X, this created a compliance gap that IAS measurement alone cannot close - because the problem is not about content adjacency but about supply chain transparency.
X's programmatic inventory generated 106 billion weekly impression opportunities, with 98% consisting of native ad formats. The average CPM for X's native inventory through programmatic channels stood at $0.39 as of December 2024 data, an unusually low floor that makes X inventory attractive to algorithmic optimization systems regardless of brand preferences. Advertisers wanting to exclude X entirely must manually add twitter.com, x.com, com.twitter.android, and the iOS app ID 333903271 to placement exclusion lists in every relevant campaign.
The IAS partnership document is carefully scoped. It lists product launches and market expansions chronologically. It does not validate X's self-reported brand safety scores. It does not confirm that 99% brand safety rates hold across all surfaces - only that IAS measurement exists on the surfaces it specifies. The distinction matters because X's playbook aggregates data from multiple sources, including X's own measurements, when citing headline safety figures.
A DoubleVerify study published in November 2025, surveying 22,000 consumers and 1,970 marketers across 21 countries, found that nearly two-thirds of social media advertisers expressed brand suitability concerns. The IAS 2026 Industry Pulse Report, released in December 2025 and covered by PPC Land, found that 53% of surveyed U.S. digital media experts cited ad adjacency to unsuitable content as a top challenge for 2026 across social platforms broadly. These figures were not specific to X, but the anxieties they describe apply with particular force to a platform that has spent three years rebuilding advertiser trust after a documented departure crisis.
The role IAS plays in that rebuilding effort is technically specific, geographically bounded, and limited in scope to the surfaces and markets listed in the partnership document. That precision is the product's value. X needs measurement that is independent, verifiable, and consistent with the standards applied elsewhere - and IAS, with its December 2025 video blocking solution and its AI-powered agent interface launched the same month, continues expanding the technical infrastructure advertisers use to make placement decisions across all major social platforms.
Whether the current IAS coverage map on X - timeline and profiles in six markets post-bid, Vertical Video in the United States pre-bid - is sufficient for the level of brand suitability assurance brands require depends on how individual advertisers weigh their tolerance for residual risk against the platform's claimed performance returns. X users are 45% more likely to buy products they have seen advertised on the platform compared to non-X users, according to GWI data from Q2 2024 through Q1 2025 cited in the playbook. X ranks first for ad attentiveness among social media platforms, per eMarketer data from April 2024. These figures appear alongside the brand safety stack in X's pitch - not as distractions from safety concerns, but as the commercial case that safety concerns need to be weighed against.
Who: Integral Ad Science (Nasdaq: IAS), a global media measurement and optimization company, in a long-running partnership with X (formerly Twitter), operating under Elon Musk's xAI umbrella. The audience for both the IAS document and X's Brand Suitability Playbook is brand advertisers, agencies, and media buyers evaluating campaign investment on the platform.
What: IAS updated its official account of its X partnership on March 2, 2026, detailing measurement and optimization milestones from 2017 through January 2026. The most recent milestone is the expansion of Brand Safety and Suitability Measurement to X profiles, adding to existing timeline feed coverage. This expansion coincides with X's formal Brand Suitability Playbook circulating to advertisers, which positions IAS measurement data - alongside DoubleVerify and X's own Grok AI - as the three-layer evidential base for claims of 99%+ brand safety and 97%+ brand suitability scores across the platform.
When: The IAS partnership document was updated on March 2, 2026. The January 2026 profile measurement expansion is the most recent technical milestone it records. The X Brand Suitability Playbook was obtained and reported by Adweek on March 3, 2026. The broader context spans the period from October 2022 - when Musk's acquisition triggered an advertiser exodus - through early 2026.
Where: Post-bid Brand Safety and Suitability Measurement from IAS covers X's timeline feed in the United States, Australia, Canada, New Zealand, the United Kingdom, and Japan, and X profiles globally as of January 2026. Pre-bid IAS Optimization applies only to Vertical Video (Immersive Video format) in the United States. Pre-bid coverage does not yet extend to the timeline, profile, or search surfaces more broadly.
Why: X's advertising revenues declined from $2.43 billion in 2021 to an estimated $1.25 billion in 2025, a gap the platform is actively working to close. Third-party measurement from IAS provides the independent credibility that X's self-reported figures alone cannot supply. For advertisers, the IAS coverage map defines the boundaries of what can be independently verified on the platform - and where first-party controls from X and Grok's AI scoring must be taken on trust.
