Why this page exists.
A data clean room is a piece of technical infrastructure, but the obligations that bind it are not. They sit in three places: the standards layer that the digital advertising industry is building from the inside, the European regulatory regime that has been the global pacesetter for privacy and data-sharing law for nearly a decade, and the United States regulatory posture that operates through a patchwork of state statutes and federal enforcement.
DCRI's view is that practitioners operating clean rooms in 2026 cannot reason about the category without holding all three layers in mind at once. The standards layer tells you what the infrastructure is converging on. The regulatory layers tell you what it is allowed to do. The gaps between them — where the standards anticipate uses the regulators have not yet addressed, and where the regulators impose obligations the standards do not yet handle — are where the most consequential decisions in this category will be made over the next 24 months.
This page maps the three layers as they stand in May 2026. It does not interpret them. Interpretation belongs to qualified counsel in the relevant jurisdiction; DCRI is an institute of practice, not a law firm. What this page provides is the inventory and the connective tissue.
The standards layer.
The primary standards body for data clean rooms in the digital advertising and marketing context is the IAB Tech Lab, whose Rearc Addressability Working Group has shipped three foundational artifacts in the past 24 months.
Data Clean Room Guidance and Recommended Practices, version 1.0. Originally released for public comment in February 2023; version 1.0 finalized in July 2024. The document establishes common principles, functions, and privacy-enhancing technologies for data clean rooms, and outlines limitations and guardrails when engaging with DCR providers. It is the closest the industry has to an agreed definition of what a data clean room is and is not. The Tech Lab maintains the canonical version at iabtechlab.com/datacleanrooms.
PAIR — Publisher Advertiser Identity Reconciliation. Donated to IAB Tech Lab by Google's Display & Video 360 team in October 2022; opened for public comment as version 1.0 in September 2024; finalized as version 1.0 in February 2025; updated to version 1.1 on July 16, 2025. PAIR is a cryptographic standard for first-party data matching between advertisers and publishers, using commutative encryption to allow secure matching of multiple-encrypted keys without exposing personal information. Version 1.1 added encoding clarifications and an Open PAIR prebid module for industry adoption. PAIR is the protocol that allows different clean rooms to talk to each other for identity matching; it is the closest thing the category has to interoperability infrastructure.
ADMaP — Attribution Data Matching Protocol. Released for public comment on October 15, 2024; comment period closed November 14, 2024; finalized as version 1.0 in February 2025. ADMaP is a protocol for secure measurement of conversion data between advertisers and publishers, using Private Set Intersection and Trusted Execution Environments. Where PAIR handles identity reconciliation, ADMaP handles attribution measurement. The two together form the spine of IAB Tech Lab's interoperability framework for the category.
What the standards layer does not cover. The IAB Tech Lab artifacts are scoped to digital advertising. They do not address healthcare, financial services, or AI/ML training-data uses of clean room infrastructure. Other standards bodies — including ISO/IEC for privacy engineering and NIST for privacy framework guidance — have produced relevant horizontal work, but no body has yet produced clean-room-specific standards for non-advertising use cases. This is an open territory.
The European regulatory layer.
The European Union operates the world's most developed regime for personal data, B2B data sharing, and the regulation of large digital platforms. Five instruments are most material to clean room practitioners.
The General Data Protection Regulation (GDPR), in force since May 25, 2018. The floor. Every other instrument in this list supplements rather than replaces GDPR. For clean room work, the most consequential GDPR concepts are lawful basis (Article 6), purpose limitation, data minimisation, the special-category data regime (Article 9), and the joint controller / processor distinction (Articles 26 and 28). Clean rooms do not provide automatic GDPR compliance; they provide a technical surface on which compliance can be architected. The institute's working position is that this distinction is the single most frequently misstated point in vendor marketing material in the category.
The Digital Markets Act (DMA), Regulation (EU) 2022/1925. Entered into force November 1, 2022; full compliance required from designated gatekeepers as of March 6, 2024. Currently designates seven gatekeepers (Alphabet, Amazon, Apple, ByteDance, Meta, Microsoft, and Booking.com) across 24 core platform services. For clean room practitioners, the most consequential provisions are Article 5(2) on the combination of personal data across services without consent and Article 6 on data portability and access for business users. Enforcement has been active: the Commission fined Apple €500 million and Meta €200 million in April 2025 for anti-steering and "pay or consent" breaches, and Google €2.95 billion in September 2025 for breaches of advertising-related antitrust rules.
The Digital Services Act (DSA), Regulation (EU) 2022/2065. Companion instrument to the DMA, focused on intermediary liability and platform transparency rather than gatekeeper data practices. Material to clean rooms primarily where they sit inside walled gardens whose hosts are designated Very Large Online Platforms or Very Large Online Search Engines.
The Data Act, Regulation (EU) 2023/2854. Entered into force January 11, 2024; core provisions applicable from September 12, 2025; further provisions phased in through September 12, 2027. The Data Act is the EU's horizontal framework for B2B data sharing and for the obligations of cloud and edge service providers. For the clean room category, the most material provisions are the Chapter II/III obligations on fair, reasonable, and non-discriminatory (FRAND) B2B data-sharing contracts and the Chapter VI obligations on cloud-provider switching and interoperability. The latter directly bear on the warehouse-native architecture (Snowflake, AWS, Databricks, BigQuery) and on the contractual terms under which a clean room operator can lock in or release a customer's data. Manufacturers of connected products and providers of related services placed on the EU market after September 12, 2026 must additionally design those products to allow user access to the data they generate.
The AI Act, Regulation (EU) 2024/1689. Entered into force August 1, 2024; phased application. Prohibitions on unacceptable-risk AI practices applied from February 2, 2025. Governance provisions and general-purpose AI model obligations applied from August 2, 2025. High-risk AI system obligations were originally scheduled for August 2, 2026, but the Digital Omnibus on AI — proposed by the Commission on November 19, 2025 and politically agreed on May 7, 2026 — defers most high-risk obligations to December 2, 2027, with product-embedded high-risk systems deferred to August 2, 2028. The AI Act is the regulation most underspecified for the clean-room-as-training-substrate pattern that has emerged with AWS Clean Rooms ML, Snowflake's collaborative training features, and Databricks Clean Rooms. The Act addresses training data provenance and governance for high-risk systems but does not directly address the specific architectural question of synthetic-data generation inside a clean room as a privacy-enhancing input to model training.
The United States regulatory layer.
The United States has no federal omnibus privacy law. Practitioners operate against a patchwork of state statutes and federal enforcement under existing authorities — primarily the FTC's Section 5 powers over unfair and deceptive practices.
The state privacy law patchwork. As of May 2026, the operative state privacy regimes most material to clean room work are the California Consumer Privacy Act (CCPA) as amended by the California Privacy Rights Act (CPRA), the Virginia Consumer Data Protection Act (VCDPA), the Colorado Privacy Act (CPA), the Connecticut Data Privacy Act (CTDPA), the Utah Consumer Privacy Act (UCPA), and the further laws adopted in Texas, Oregon, Montana, Iowa, Indiana, Tennessee, Delaware, New Hampshire, New Jersey, Maryland, Minnesota, Rhode Island, and others. The laws are convergent in structure but differ materially on sensitive-data definitions, opt-out mechanics, and the treatment of "sharing" versus "sale." Clean room practitioners operating across the US should not assume that a single privacy posture satisfies all applicable state regimes.
The FTC's position on data clean rooms. In November 2024, the FTC's Office of Technology published Data Clean Rooms: Separating Fact from Fiction, which is the single most consequential US regulatory statement on the category to date. The FTC's working position is that liability for violations of the FTC Act is not mitigated by clean room infrastructure, and that the use of a clean room does not insulate a company from Section 5 liability if the underlying data sharing would otherwise be unlawful. Practitioners should treat this position as the operative federal view in the absence of new federal legislation.
The FTC's enforcement posture on data brokers and location data. The Commission has brought a series of enforcement actions against data aggregators and location data brokers under Section 5 of the FTC Act: Kochava (August 2022), X-Mode / Outlogic (January 2024 consent order), InMarket Media (January 2024 consent order), Mobilewalla (December 2024 proposed order; finalized January 2025), and Gravy Analytics / Venntel (December 2024). The Mobilewalla order is notable as the first FTC action prohibiting collection of consumer data from real-time bidding exchanges for purposes other than participating in those auctions — a structural prohibition that bears on the line between RTB participation and downstream data-broker activity. Clean room operators that ingest data from RTB exchanges or from data-broker pipelines should treat this enforcement line as directly applicable.
Sectoral regimes. HIPAA and the HITECH Act govern healthcare data; the Gramm-Leach-Bliley Act and Regulation S-P govern financial-services customer data; FERPA governs educational records; COPPA governs data on under-13 users. Each imposes its own obligations on what data can be brought into a clean room collaboration and under what conditions. DCRI's v1 scope is advertising and marketing DCRs; healthcare and financial services follow in later releases.
The connective tissue — where the layers do not yet meet.
The most consequential decisions in this category over the next 24 months will be made in the seams between the three layers above. DCRI's working list of the most underspecified seams as of May 2026:
PAIR and lawful basis under GDPR. PAIR specifies the cryptographic protocol for identity reconciliation but does not specify which lawful basis under GDPR Article 6 a controller should rely on for the underlying data processing. The protocol is content-neutral on legal basis by design; the obligation to identify a valid basis remains with the data controllers. The practical question — whether consent or legitimate interests is the more defensible basis for PAIR-based matching in EU contexts — is not addressed in the IAB Tech Lab artifact and is open to disagreement among practitioners.
ADMaP and the DMA's gatekeeper data obligations. ADMaP standardises attribution measurement between advertisers and publishers, but does not address whether the same protocol satisfies a gatekeeper's obligations under DMA Article 6 to provide business users with access to data generated through their use of the gatekeeper's platform. The institute's working position is that ADMaP as currently specified is necessary but not sufficient for DMA Article 6 compliance.
The AI Act and the clean-room-as-training-substrate pattern. The AI Act's provisions on training data provenance for high-risk systems do not directly address the specific architectural question of whether synthetic data generated inside a clean room — for example, via AWS Clean Rooms ML's synthetic-data generation feature — qualifies as appropriately governed training data under the Act. The deferral of high-risk obligations to December 2027 widens the window in which this gap will remain unresolved at the legislative level.
The Data Act's FRAND obligations and warehouse-native clean room contracts. The Data Act's Chapter II/III obligations on fair, reasonable, and non-discriminatory B2B data-sharing terms enter into application for new contracts from September 12, 2025 and for legacy contracts from September 12, 2027. The provisions bear directly on the contractual terms under which a warehouse-native clean room operator (Snowflake, AWS, Databricks, BigQuery) makes its services available to EU-based customers. The institute is not aware of public guidance from any cloud provider as to how the Data Act's switching and interoperability obligations are being implemented at the clean-room product layer specifically.
State-level versus federal posture in the United States. The FTC's November 2024 Office of Technology statement is the most consequential federal regulatory voice on the category, but the FTC's enforcement docket and the state privacy regimes can diverge on what counts as a permissible use of clean room infrastructure. The Mobilewalla order's structural prohibition on RTB-data collection for downstream purposes other than auction participation is a federal posture; the state regimes have not converged on an equivalent rule.
What this page is not.
This page is not legal advice and is not a substitute for qualified counsel in any jurisdiction. It is also not a substitute for the primary sources it points to: the IAB Tech Lab specifications, the EUR-Lex texts of the regulations cited, the FTC's published statements, and the state-level statutes themselves. Practitioners should consult the primary sources before making operational or contractual decisions.
This page is also not exhaustive. The institute's working list of regulatory regimes material to clean room practice in 2026 includes the UK's post-Brexit divergence (the UK GDPR and the Data Use and Access Act 2025), the EDPB's published guidance on data processors and joint controllers, Brazil's LGPD, the regimes of the Gulf Cooperation Council states, India's DPDP Act, and the developing AI governance regimes in Canada, South Korea, and Japan. DCRI will extend the map over time as practitioner contributions sharpen the institute's coverage.
Contributing.
DCRI's standards-literacy claim depends on the accuracy of this page. The institute invites contributions from practitioners working in the standards layer, the EU regulatory layer, and the US enforcement layer. Material errors should be reported to info@dcrinstitute.org with a citation to the primary source. Substantive expansion of the connective-tissue section — naming additional seams where the layers do not meet — is the contribution the institute most values.
v0.9 · Last updated: May 2026. The institute commits to a quarterly review cadence for this page; readers may treat the dateModified timestamp as authoritative.