Providers of very large online platforms or of very large online search engines shall compile and make publicly available in a specific section of their online interface, through a searchable and reliable tool that allows multicriteria queries and through application programming interfaces, a repository containing the information referred to in Article 26(3) for each advertisement displayed on their online interfaces.
The repository shall be easily accessible for browsing and shall be accessible to the Digital Services Coordinators and the Commission, without requiring registration or identification.
The information in the repository shall be kept for a period of one year after the last time the advertisement was displayed. The repository shall be updated daily.
The information in the repository shall not contain any personal data of the recipients of the service to whom the advertisement was or could have been presented.
Understanding This Article
Article 41 represents a transformative shift in online advertising accountability by mandating comprehensive public transparency of all advertising on Very Large Online Platforms and Very Large Online Search Engines. While Article 26 requires platforms to show individual users transparency information about specific advertisements they encounter, Article 41 creates systemic, aggregate transparency: a complete public archive of all advertisements enabling researchers, journalists, civil society organizations, regulators, and the general public to analyze advertising patterns, targeting practices, spending, and content at scale across entire platform ecosystems.
The repository requirement serves multiple critical accountability functions that address fundamental democratic and economic concerns. First, election integrity and democratic accountability: the public can scrutinize political advertising during election campaigns, identifying who is spending how much, what messages target which audiences, whether foreign actors or undisclosed organizations are buying political ads, and whether platforms enforce political advertising policies consistently. This addresses concerns about 'dark money' political advertising, foreign interference in elections, and the lack of transparency that characterized the Cambridge Analytica scandal and subsequent election interference incidents. Second, commercial transparency and consumer protection: consumers and competitors can monitor advertising practices, pricing strategies, targeting approaches, and potentially deceptive advertising. This enables competition authorities to detect anti-competitive behavior, consumer protection agencies to identify fraudulent advertising, and consumers to understand how they are being targeted. Third, research enablement: academics and journalists can conduct systematic studies of advertising ecosystems, misinformation spread through paid promotion, discriminatory ad targeting in housing/employment/credit, child-directed marketing, health misinformation advertising, and platform content moderation effectiveness. Fourth, regulatory oversight and proactive supervision: authorities can monitor compliance with advertising rules, identify systemic problematic patterns, detect illegal advertisements, and prepare for audits or enforcement proceedings with comprehensive data. Fifth, public accountability and informed discourse: platform advertising systems become transparent rather than proprietary black boxes, enabling informed public and policy debate about platform business models, the attention economy, and the societal implications of targeted advertising.
The repository must contain comprehensive information specified in Article 26(3) for each advertisement: (a) clear indication that content is advertising; (b) natural or legal person on whose behalf advertisement is presented (advertiser identity with verification status); (c) natural or legal person who paid for advertisement if different from advertiser (revealing hidden sponsorship); (d) meaningful information about main parameters determining which recipients of service see advertisement (targeting criteria such as age ranges, geographic locations, interests, behaviors); (e) where applicable, information about advertisement targeting based on profiling and specific parameters used for profiling (algorithmic targeting transparency); (f) period during which advertisement was presented (start and end dates); (g) total number of recipients of service reached or estimated to have been reached (impression counts, often in ranges to balance transparency with commercial sensitivity). This comprehensive metadata enables understanding not just advertisement content but the complete advertising transaction: who paid for it, who they targeted, how they targeted them, and how widely the advertisement spread.
Critical technical and usability requirements ensure repository practical utility rather than nominal compliance. First, 'searchable and reliable tool' means not merely a data dump but a functional, well-designed search interface that users can actually navigate. Users should be able to query by advertiser name, keywords in advertisement content, topic categories, date ranges, number of impressions, targeting parameters, and platforms where shown. The tool must be reliable with consistent uptime, reasonable performance, and accurate results. Second, 'multicriteria queries' requires advanced search functionality combining multiple filters simultaneously (e.g., 'political advertisements mentioning climate change, targeted to users aged 18-35 in Germany, displayed during October 2024, with over 1 million impressions'). This prevents platforms from limiting transparency through artificially constrained search capabilities. Third, 'application programming interfaces' (APIs) mandate machine-readable data access enabling researchers and organizations to conduct large-scale systematic analysis programmatically. API access prevents platforms from using poorly-designed user interfaces to make data technically public but practically inaccessible. APIs must be well-documented, reliable, have reasonable rate limits balancing platform resources with research needs, and provide structured data in standard formats. Fourth, 'easily accessible for browsing' requires repository to be prominently available, not buried in obscure website sections or hidden behind multiple navigation layers. Fifth, 'without requiring registration or identification' ensures completely barrier-free public access. Anyone can access the repository including users outside the EU, non-registered users, journalists without platform accounts, researchers from any institution, and civil society organizations globally.
The one-year retention period balances competing considerations. It ensures sufficient historical data for retrospective analysis (e.g., analyzing election advertising 6 months after an election to study its impact, examining advertising pattern evolution over time, identifying coordinated campaigns) while not requiring permanent archiving which could impose excessive storage costs and create privacy concerns about indefinite data retention. Some platforms voluntarily exceed this minimum, with Meta retaining political advertisements for 7 years. Daily updates ensure the repository reflects current advertising activity, not outdated snapshots, enabling near-real-time monitoring of advertising patterns, rapid response to problematic advertising trends, and timely analysis of emerging issues. The prohibition on personal data about advertisement recipients prevents privacy violations: repositories show aggregate advertisement information (advertiser, content, targeting parameters, total impressions) but not individual user information (which specific users saw an advertisement, when they saw it, whether they clicked, their subsequent behavior). This balances transparency about advertisers and targeting with privacy protection for users, complying with GDPR while enabling meaningful advertising ecosystem analysis.
Article 41's public repository requirement distinguishes the DSA from most advertising regulation globally, which focuses on what advertisements can say or how they can target but doesn't require systemic transparency. By mandating comprehensive public advertisement archives, Article 41 enables 'transparency-based governance' where sunlight serves as a disinfectant. Advertisers knowing their advertisements will be publicly documented may voluntarily avoid problematic practices to prevent reputational damage. Platforms implementing discriminatory advertisement systems face public exposure and scrutiny. Election manipulation through opaque advertising becomes visible to watchdogs, journalists, and researchers. This transparency creates accountability pressure operating continuously, not just during enforcement actions.
Compliance verification occurs through Article 37 independent audits and ongoing regulatory monitoring. Auditors examine: Is the repository complete, including all advertisements displayed on the platform without exclusions? Is search functionality robust, allowing users to effectively find relevant advertisements? Are APIs functional, well-documented, and providing comprehensive data access? Is metadata accurate, complete, and properly categorized? Are retention obligations met with advertisements remaining accessible for the full one-year period? Are update obligations met with daily refreshes? Auditors may compare repositories against independent advertisement monitoring services, crowdsourced advertisement collection, or the platform's own advertising data to verify completeness. Platforms face ongoing pressure to maintain and improve repositories as researchers and civil society organizations identify gaps, functionality problems, or compliance shortcomings.
Practical Application
For Meta (Facebook and Instagram Ad Library Implementation): Meta has established the Meta Ad Library as its Article 41 repository covering Facebook and Instagram advertising globally, with enhanced features for EU compliance. Implementation elements: (1) Accessibility: Dedicated URL (facebook.com/ads/library) prominently linked from main platform footers and transparency centers. No login required; completely public access from any device or location. Mobile-optimized interface for smartphone access. (2) Search functionality supports: Advertiser name search ('search all advertisements from Nike' showing all current and historical ads from that advertiser), keyword search ('advertisements mentioning vaccines' finding ads containing specific terms), date range filtering (show ads displayed between specific dates), country/region selection (filter to advertisements shown in specific Member States), advertisement category filtering (political/social issue advertisements vs. all advertisements, with political ads having enhanced transparency). Multicriteria queries combine filters ('political advertisements in Germany mentioning climate change, January-March 2024' returning specific subset). (3) Advertisement entry display: Complete advertisement content (text, images, video, links), Advertiser name with verification status badge, Platform(s) where shown (Facebook, Instagram, Messenger, Audience Network), Start date and status (active/inactive), Impressions in ranges ('10,000-50,000' balancing transparency with commercial sensitivity), Spend in ranges ('€1,000-5,000'), Detailed targeting information including age ranges, locations, and for political advertisements full detailed targeting parameters like interests and behaviors. (4) API provides programmatic research access: Researchers with approved access can query entire advertisement database, filter by multiple criteria, download bulk datasets, conduct systematic longitudinal studies. Meta documents API endpoints, provides code examples, manages rate limits to prevent abuse while enabling legitimate research. (5) Daily updates: New advertisements appear within 24 hours of launch; discontinued advertisements update status daily. (6) Retention: One-year retention for general advertisements; seven-year retention for political/social issue advertisements, voluntarily exceeding legal requirements to support election research and democratic accountability. Compliance challenges Meta addresses: Ensuring complete coverage (all advertisements included without exclusions or technical gaps), accuracy of metadata (particularly targeting information which must reflect actual targeting used), search quality (users can find relevant advertisements without false negatives), API reliability (consistent uptime, reasonable rate limits, complete documentation), political advertisement categorization (correctly identifying political/social issue advertisements requiring enhanced transparency), advertiser verification (confirming advertiser identity and authorization to run advertisements). Audit verification mechanisms: Independent researchers compare Ad Library against monitoring tools tracking Facebook advertisements through browser extensions and crowdsourcing, test search functionality with known advertisement queries, verify API completeness and accuracy, assess metadata quality against ground truth, examine retention periods. Meta must continuously maintain and improve Ad Library technical infrastructure, address gaps and problems identified through research and audits, respond to regulator inquiries, and adapt to evolving transparency expectations.
For Google (YouTube, Search, Display Network Ads Transparency Center): Google operates the Ads Transparency Center as its Article 41 repository covering YouTube, Google Search, Google Display Network, and other Google advertising services. Implementation: (1) Dedicated transparency.google.com website with comprehensive advertisement search functionality accessible without Google account. (2) Search capabilities: Advertiser name (find all advertisements from specific advertiser), advertiser verification ID (unique identifier for verified advertisers), advertisement format filtering (video advertisements on YouTube, text search advertisements, display banner advertisements, shopping advertisements), date range selection, region targeting (which countries/regions saw advertisements), topic filtering for political/election advertisements. API access for approved researchers conducting systematic studies. (3) Advertisement display presentation: Creative preview (advertisement content shown in platform-appropriate format: video player for YouTube ads, text preview for search ads, image for display ads), Advertiser identity and verification status, Advertisement formats used (which Google services showed the ad), Regions where targeted and shown, First shown and last shown dates, Impressions (in ranges), For political advertisements: additional required disclosures including payer identity if different from advertiser, issue categories (which political topics), detailed targeting parameters. (4) Daily updates across Google's advertising ecosystem with automated synchronization. (5) One-year retention for general advertisements; extended retention for political advertisements supporting election research. Compliance considerations unique to Google: Google's advertising ecosystem spans multiple diverse services (Search text advertisements, YouTube video advertisements, Display Network banner/rich media advertisements, Shopping product advertisements) requiring unified repository across fundamentally different advertisement formats and platforms. This creates technical challenges in creating consistent metadata and display across advertisement types. Challenge: Search advertisements are primarily text-based with limited creative content, while YouTube advertisements are full video productions, and display advertisements are visual banners, requiring repository to accommodate diverse formats in searchable, comparable ways. API must provide structured data that works across varied formats enabling cross-platform analysis. Google must ensure advertisers cannot circumvent transparency by misclassifying political advertisements as commercial, using shell companies or complex corporate structures to hide true sponsors, or exploiting differences between services' verification requirements. Verification and categorization accuracy is critical for repository utility. Audit focus areas: Completeness across Google's multiple advertising networks ensuring advertisements on Search, YouTube, Display, Shopping all appear; verification system effectiveness in identifying true advertisers; search and API functionality enabling researchers to find relevant advertisements across formats; political advertisement categorization accuracy using consistent criteria; metadata accuracy particularly for targeting parameters which differ across Google's services.
For TikTok (Developing and Maintaining TikTok Ad Library): TikTok is establishing and continuously improving its Advertisement Library for Article 41 compliance as a designated VLOP. Implementation requirements and challenges: (1) Dedicated advertisement library website (ads.tiktok.com or integrated section within TikTok.com) accessible without login, with prominent placement discoverable from main application. (2) Search functionality adapted to TikTok's video-first, music-driven, hashtag-centric advertising ecosystem: advertiser name search, keywords in advertisement content (including video captions, on-screen text, audio transcripts), music/sounds used in advertisements (enabling research on music licensing in advertising, popular audio trends), hashtag search (critical for TikTok's hashtag-driven discovery and branded hashtag challenges), date ranges, geographic regions, age targeting (particularly important given concerns about TikTok's young user base), impressions ranges. Given TikTok's short-form video format, search should support video content analysis including captions, audio, visual elements, effects used. (3) Advertisement display: Video preview or full video playback (requiring substantial video hosting infrastructure), Advertiser name and verification status, Targeting parameters used (age, gender, location, interests, device types, user behaviors), Display period (start/end dates), Impression counts (total views), Link destinations or call-to-action, For political advertisements: enhanced disclosures. (4) API providing: Video download or streaming access for research purposes with appropriate use restrictions preventing unauthorized redistribution, metadata in structured format (JSON/XML) with comprehensive fields, bulk query capabilities for large-scale research, documentation and code examples. (5) Daily updates reflecting new advertisements and status changes, one-year retention. Unique challenges TikTok faces: (1) Video storage and streaming infrastructure: Repository must host and serve potentially millions of video advertisements, requiring substantial cloud infrastructure, content delivery networks, video encoding/transcoding, and bandwidth. Unlike text or image advertisements, video imposes much higher technical requirements. (2) Music licensing and copyright: TikTok advertisements frequently use popular music, and archiving these advertisements publicly may face music licensing restrictions. TikTok must negotiate rights or implement technical solutions like audio filtering, legal disclosures, or licensing extensions covering repository use. (3) International advertiser verification: TikTok's global reach includes advertisers from diverse jurisdictions including those with limited identity verification infrastructure or different corporate transparency norms. Ensuring advertiser identity accuracy across regions (China, Southeast Asia, Latin America, Europe) with varying verification standards is complex. TikTok must develop verification procedures working across regulatory environments. (4) Native advertisement format identification: TikTok's native advertising formats including Branded Hashtag Challenges (where users create content with sponsored hashtags), Creator Marketplace (sponsored creator content), In-Feed Ads (appearing in main content feed), TopView (full-screen ads on app open), Brand Takeover blur lines between organic content and commercial communication. Repository must clearly identify all commercial communications including these native formats, ensuring users understand what is advertising. (5) Youth protection considerations: Given TikTok's young user demographic and regulatory scrutiny about youth protection, advertisement repository enables monitoring of advertising targeting minors, age-inappropriate content, products prohibited from youth marketing (vaping, alcohol, gambling, diet products), potentially exploitative advertising practices. This transparency creates accountability pressure. Audit verification priorities: Test video playback functionality across devices and networks, verify inclusion of all advertisement formats including native advertising and branded content, assess advertiser verification accuracy particularly for international advertisers, test API functionality and download capabilities, examine youth protection by analyzing advertisements targeted to young users, evaluate search effectiveness for TikTok's unique video/music/hashtag ecosystem.
For Researchers (Using Advertisement Repositories for Public Interest Research): Article 41 repositories enable independent research impossible under pre-DSA opacity when platform advertising operated as proprietary black boxes. Research applications with real-world examples: (1) Election integrity and political advertising studies: Analyze political advertising spending, messaging strategies, targeting tactics, and potential foreign interference during election campaigns. Example: Researcher queries all political advertisements mentioning specific candidate in final week before election, analyzes spending by supporter vs. opponent groups, identifies different messages targeted to different demographic groups, documents potential foreign-funded advertisements, compares advertisement content across platforms, examines micro-targeting of swing voter groups. Findings can inform public understanding of campaign strategies, regulatory enforcement of election laws, and policy debates about political advertising regulation. (2) Misinformation and disinformation tracking: Identify advertisements promoting false health claims, scams, conspiracy theories, manipulated media, or coordinated disinformation campaigns. Cross-reference advertisement content with fact-checking databases, identify repeat offender advertisers evading platform policies through repeated account creation, track evolution of misleading claims across advertisement campaigns. Example: During COVID-19 pandemic, researchers documented advertisements promoting unproven treatments, anti-vaccine misinformation, and scam medical products, providing evidence for platform policy enforcement and regulatory action. (3) Discrimination analysis in housing, employment, and credit advertising: Examine whether housing, employment, credit, and insurance advertisements exhibit discriminatory targeting patterns excluding protected groups. Example: If job advertisements systematically exclude older workers from target audiences, housing advertisements exclude minority neighborhoods from geographic targeting, or credit advertisements show primarily to men rather than women with similar financial profiles, repository provides documentary evidence for discrimination claims and civil rights enforcement. This research has led to settlement agreements and platform policy reforms. (4) Issue advocacy and astroturfing monitoring: Track corporate and advocacy group spending on issue campaigns (climate change, healthcare reform, immigration, regulatory changes). Analyze message framing, identify apparent grassroots movements actually funded by undisclosed corporate interests (astroturfing), document coordination among seemingly independent groups running similar advertisements. Example: Environmental researchers track fossil fuel industry advertising during climate policy debates, revealing spending amounts, messaging strategies, and front groups. (5) Platform comparison and policy effectiveness: Compare advertising practices, transparency, and content moderation across platforms. Does TikTok permit health misinformation advertisements that Facebook rejects? Do political advertisement targeting options differ between platforms? Are verification requirements enforced consistently? Cross-platform analysis reveals policy gaps and best practices. Researchers use APIs to download bulk data, apply natural language processing to analyze advertisement content themes and sentiment, use computer vision to analyze image/video advertisements, apply network analysis to identify coordinated campaigns, analyze targeting patterns using statistical methods, track longitudinal trends over time. Repository transparency converts platform advertising from proprietary black box to public infrastructure for accountability research. Limitations researchers must address: Repositories show advertisements that ran, not advertisements platforms rejected (cannot fully analyze moderation effectiveness or submission patterns); impression counts often in ranges rather than exact numbers (limits precision of reach analysis); targeting parameters shown may not capture all variables platforms actually use algorithmically (potential hidden targeting factors); no data on advertisement effectiveness (clicks, conversions) which remains proprietary; cannot identify individual users who saw advertisements (appropriate privacy protection but limits micro-level analysis). Nevertheless, Article 41 repositories represent unprecedented advertising transparency enabling empirical analysis of platform advertising ecosystems previously impossible.
For Regulators (Using Repositories for Proactive Supervision and Enforcement): Digital Services Coordinators and Commission use Article 41 repositories for proactive monitoring without requiring platforms produce data for each investigation. Applications: (1) Systematic illegal advertisement monitoring: Regulators can programmatically monitor for advertisements violating specific legal prohibitions (tobacco advertisements to minors, illegal gambling, prohibited health claims, fraudulent investment schemes, counterfeits). When identified, regulators investigate why platform's own content moderation didn't prevent them, assess systemic failures, and initiate enforcement. Example: DSC identifies pattern of illegal gambling advertisements targeting young users; investigation reveals platform verification gaps enabling repeated policy violations. (2) Pattern analysis identifying systemic compliance issues: Are certain advertisers repeatedly violating policies suggesting verification failures? Are advertisements targeting minors with age-inappropriate content indicating youth protection weaknesses? Are foreign state actors buying political advertisements despite prohibitions? Pattern analysis reveals systemic problems requiring structural remediation not just individual advertisement removal. (3) Cross-platform coordination and consistency: Commission compares advertising across VLOPs identifying inconsistent enforcement. If TikTok permits advertisement category that Meta prohibits, is TikTok violating content standards or is Meta over-enforcing? Cross-platform comparison identifies best practices and problematic outliers. (4) Audit preparation and targeted investigation: Before conducting Article 37 audit, regulators analyze platform's advertisement repository identifying areas of concern to focus audit investigation, preparing specific questions, selecting advertisement samples for detailed review. Repository analysis makes audits more efficient and effective. (5) Enforcement evidence and documentation: Repository provides documented evidence for enforcement proceedings. Rather than relying solely on complaints or platform self-reporting, regulators can prove platform displayed illegal advertisements X times over Y months with Z impressions, supported by platform's own repository records. This evidence is difficult for platforms to dispute. Commission and DSCs must develop technical capacity to query and analyze repositories at scale: Repositories generate massive data volumes (millions of advertisements, complex metadata) requiring sophisticated analysis tools, not manual review. Regulators should develop automated monitoring systems flagging likely illegal advertisements for human review, pattern detection algorithms identifying systemic issues, cross-platform comparison tools enabling consistency analysis, data science capabilities for statistical analysis, partnerships with research institutions for advanced analysis. Article 41 shifts regulatory approach from reactive complaint-response to proactive transparency-enabled supervision, where regulators can identify problems before significant harm occurs and monitor compliance continuously rather than episodically.