The 'Data Clean Room' Danger: Why Your Anonymized Data Isn't Safe (And What Corporations Know That You Don't)

The corporate world is excited. Marketers, publishers, and advertisers have discovered what they're calling a privacy solution: data clean rooms. The concept is seductive. Two companies can analyze combined datasets without exposing raw data. Individual privacy is protected. Everyone wins. Regulators are satisfied. The technology is booming—the global data clean room market is projected to reach $5.6 billion by 2030. Companies are racing to implement it.

There's just one problem: the Federal Trade Commission has gone on record stating that data clean rooms "are not rooms, do not clean data, and have complicated implications for user privacy, despite their squeaky-clean name."

What the FTC is delicately describing is a crisis that most individuals don't understand: data clean rooms aren't privacy solutions—they're privacy theater masking the most sophisticated re-identification infrastructure ever created. They enable corporations to pool your personal information in ways that feel safe to regulators while making you exponentially more vulnerable to tracking, targeting, and discrimination.

This is the untold story of the data clean room industry: how a technology marketed as privacy protection is actually amplifying surveillance, how anonymization consistently fails to protect individuals, and how your information can be safely re-identified in these supposedly secure environments. More importantly: it's the story of why DisappearMe.AI is essential for protecting yourself from this emerging threat.

What Corporations Tell You About Data Clean Rooms (The Marketing Version)

Data clean rooms are described as sophisticated secure environments where multiple organizations can collaborate on sensitive data without exposing raw information. The pitch is compelling and works on multiple audiences.

The Corporate Promise: Collaboration Without Exposure

The industry narrative goes like this: in a data clean room, you upload your customer data and partner data into a secure, neutral environment. The data is encrypted and anonymized. Partners can perform agreed-upon analytics—measuring campaign effectiveness, analyzing audience overlap, running attribution models—without ever seeing each other's raw data. Insights flow out in aggregated form. Individual privacy is protected. Regulators are happy. Insights are shared.

This story appeals to corporate executives because it promises something that previously seemed impossible: unlocking value from sensitive data while remaining compliant with regulations like GDPR, CCPA, and HIPAA. It appeals to privacy advocates because theoretically, the data remains encrypted and individuals remain anonymous. It appeals to regulators because the architecture appears to include privacy protections.

The platforms implementing this promise sound legitimate and technical. Snowflake, the $65 billion cloud data company, offers clean room functionality. LiveRamp markets specialized clean room services. Decentriq promises zero-trust architecture. Oracle, Acxiom, and other major data players have launched clean room offerings. If this many sophisticated companies are implementing it, the logic goes, it must be legitimate.

The language used to describe clean rooms amplifies the illusion of safety. Terms like "privacy-enhancing technologies," "differential privacy," "secure multiparty computation," "pseudonymization," "anonymized outputs," and "encrypted processing" create an impression of rigorous technical safeguards. For corporate decision-makers who aren't computer scientists, this language is sufficiently technical to appear credible.

The Regulatory Promise: Compliance and Control

Data clean rooms are presented to regulators as tools that enable privacy compliance. The FTC's own description acknowledged that "data clean rooms can add privacy protections and address certain risks when configured properly." The implication is that with proper configuration, they're compliance solutions.

This framing has convinced many companies that implementing clean rooms helps them meet regulatory obligations. If you're using a data clean room, the thinking goes, you've addressed data sharing risks and privacy concerns. You've implemented privacy-enhancing technologies. You've done your due diligence.

The regulatory appeal is so strong that many companies are prioritizing clean room adoption as a compliance strategy. They're investing resources in integrating with clean room providers, training teams on clean room workflows, and building clean room partnerships. The assumption is that this investment in privacy technology demonstrates regulatory commitment.

What the FTC Actually Said (And Why It Matters)

In November 2024, the FTC published a blog post that shattered the clean room narrative. The message was direct, forceful, and unprecedented in its clarity. The agency flat-out stated that data clean rooms are being misused for privacy washing and that it plans to monitor this sector closely.

The Three Primary Dangers the FTC Identified

Danger #1: Configuration Vulnerability - Data clean rooms aren't inherently private. They're only as protective as their configuration. Many clean room platforms come with default settings that allow unrestricted data combination and extraction—the same privacy-violating approach that made cookies infamous. Companies using these platforms don't necessarily understand their configuration or aren't implementing restrictive privacy settings. The result is data sharing that feels protected but provides no meaningful privacy safeguards.

Danger #2: Data Quality Issues - "Data that enters a clean room is dirty," according to privacy compliance experts cited in the FTC guidance. The data being pooled in these rooms is often unconsented, inaccurate, unreliable, and potentially unlawful. Companies may not have legitimate legal bases for processing this data or may not have obtained proper consent. Putting dirty data into a clean room doesn't clean it—it just obscures the privacy violations happening upstream.

Danger #3: Obstruction of Privacy Obligations - Companies are using data clean rooms to evade their core privacy responsibilities. They believe that pooling data inside a clean room absolves them of obligations to protect that data or respect consumer rights. The FTC was explicit: using a data clean room doesn't eliminate your privacy obligations. It doesn't magically clean problematic data practices. It doesn't excuse inadequate consent or compliance procedures.

Beyond these three primary dangers, the FTC also warned that clean rooms can "increase the volume of disclosure and sale of data" and "provide a pathway for information exchange between untrusted parties." The very technology being marketed as privacy protection can amplify data sharing risks.

The Enforcement Warning

Perhaps most importantly, the FTC signaled that it plans to enforce against companies misusing clean rooms. The agency is monitoring this sector and will take action against companies that:

• Use clean rooms deceptively to claim privacy protections they don't provide
• Claim they're GDPR or CCPA compliant through clean room use when they're actually not
• Hide problematic data practices behind the technology
• Use clean rooms to evade core privacy obligations

This isn't theoretical regulatory posturing. The FTC has a track record of aggressive enforcement around deceptive privacy claims and has previously taken action against companies making false privacy promises. The agency is signaling that clean room misuse will face similar enforcement.

The Anonymization Illusion: Why "Anonymous" Data Can Be Identified

The fundamental vulnerability in data clean rooms stems from a false assumption that pervades the industry: that truly anonymous data exists and that once anonymized, data cannot be re-identified. Both assumptions are dangerously wrong.

The Anonymization Spectrum: A False Sense of Security

Privacy professionals distinguish between different levels of data protection. True anonymization is theoretically impossible. What exists instead is a spectrum:

Personal Data contains explicit identifiers (names, email addresses, phone numbers, Social Security numbers). This is clearly identifiable.

Pseudonymized Data replaces identifiers with codes or aliases (replacing "John Smith, john@example.com" with "User_12345"), but the link between the pseudonym and the real person remains. If you can connect the pseudonym back to the person, it's no longer anonymous.

Anonymized Data theoretically removes all information that could identify an individual, even with access to external data sources. True anonymization is extremely rare and difficult to achieve.

The critical legal distinction under GDPR and similar regulations is that pseudonymized data is still considered personal data. GDPR explicitly states that pseudonymized data remains subject to GDPR requirements, including having a legal basis for processing and implementing data security measures. The implication is clear: most "anonymized" data in data clean rooms is actually pseudonymized—still identifiable, still subject to privacy regulations, still exposing individuals to privacy risks.

The industry's marketing often conflates these categories. Companies claim their data is "anonymized" when it's actually "pseudonymized." They assert that their clean room data outputs are "anonymous" when they're actually just coded differently. This linguistic substitution creates a false sense of security.

How Re-Identification Actually Works in Practice

Re-identification—the ability to connect supposedly anonymous data back to specific individuals—is far easier than most companies realize. Sophisticated attackers and even sophisticated algorithms can often re-identify individuals from supposedly anonymized datasets through several mechanisms:

Linking with Public Data - If a dataset contains ZIP code, gender, and age, these three data points alone can uniquely identify 87% of the U.S. population. Combine that supposedly anonymous data with publicly available information (voter registration showing everyone's age and gender by ZIP code, real estate records showing property owners by ZIP code), and re-identification becomes trivial.

Behavioral Pattern Matching - If a dataset contains timestamps and behavioral sequences (website visits in a specific order, purchases of unusual item combinations, search patterns), these behavioral patterns are essentially unique fingerprints. They can often be matched to specific individuals by analyzing their known digital behavior.

Attribute Inference - Modern machine learning can infer sensitive attributes from seemingly innocuous data. Given shopping patterns, algorithm models can infer age, political affiliation, sexual orientation, health conditions, and financial status with unsettling accuracy. These inferred attributes can then be used for re-identification.

Linkage Attacks - If multiple supposedly separate datasets can be linked (matching records across databases based on partial identifiers or inferred relationships), the combination of linked datasets can expose individual identities. The more datasets that are combined in a clean room, the more linkage opportunities emerge.

Data Correlation with External Sources - Increasingly sophisticated attackers maintain external databases of individuals' behavioral and attribute information. By correlating supposedly anonymous data from a clean room with these external databases, they can identify who specific records represent.

A famous case illustrates the vulnerability: researchers were able to re-identify individuals in a de-identified dataset of healthcare records from Washington state using only age, gender, and ZIP code matched against publicly available voter registration data. The dataset was supposed to be anonymized. It wasn't.

The Current State of Anonymization Technology in Clean Rooms

Most data clean rooms use one of three technical approaches to create the illusion of anonymization:

Differential Privacy adds carefully calibrated "noise" to datasets to hide individual data points. However, the amount of noise required to truly protect privacy typically reduces the utility of the data for analysis. The more useful the analysis, the less privacy is actually protected. Companies often dial down the privacy protections to maintain data utility, recreating the original privacy vulnerabilities.

Secure Multiparty Computation (SMPC) allows multiple parties to jointly analyze data while keeping inputs private. However, SMPC protects the computation process, not the data itself. Once data enters the computation and outputs emerge, privacy depends on how restrictive output controls are. Many clean rooms allow unrestricted output extraction that reintroduces all the original privacy risks.

Encryption-in-Use encrypts data while it's being processed, preventing even the system operators from viewing raw data. However, if adversaries can observe the computations being performed and their results, they can often infer the underlying data. Encryption-in-use is theoretically stronger, but most clean room implementations use simpler approaches.

The practical reality is that most data clean rooms implement weak versions of these technologies. They prioritize data utility over privacy protection. Companies want results detailed enough to be actionable, which means they disable or weaken the privacy protections that would make the data truly anonymous.

The Corporate Reality: How Data Clean Rooms Actually Work

Behind the marketing language is a far darker reality about what's actually happening in these supposedly neutral, privacy-protective environments.

The Hidden Architecture: Sophisticated Re-Identification Infrastructure

Data clean rooms, in practice, are sophisticated re-identification infrastructure disguised as privacy-protective technology. Their actual function—beneath the privacy-preserving rhetoric—is enabling corporations to pool data from multiple sources while circumventing traditional privacy restrictions.

Here's what actually happens: Company A uploads customer data including names, emails, purchase history, and behavioral data to a clean room. Company B uploads partner data including mobile advertising IDs, location information, and device identifiers. The clean room provider implements algorithms to match records across these datasets, connecting the supposedly separate customer databases.

Once records are matched, analysis happens inside the "secure" clean room environment. But here's the key: the matching itself is the re-identification. The moment Company A's customer database is linked to Company B's advertising database, individuals are identified across previously separate systems. The clean room hasn't protected privacy—it's enabled identity linking at massive scale.

The outputs extracted from these environments are described as "aggregated" or "anonymized," but in reality, they often contain enough detail to re-identify individuals or at least identify sufficiently specific segments that individuals within those segments can be uniquely targeted. A result like "females aged 34-36 in ZIP code 94301 who purchased three specific beauty products in the past month" may technically be anonymized, but it's specific enough to uniquely identify individuals within the segment.

The Data Quality Problem: Dirty Data Gets "Cleaned" But Stays Dirty

A critical vulnerability in the clean room model is the "dirty data" problem. Much of the data entering clean rooms is:

• Unconsented - Collected without individuals' knowledge or permission, sometimes through deceptive practices
• Inaccurate - Containing errors, duplicates, and mismatches that aren't corrected before processing
• Legally Problematic - Subject to disputes about whether companies had legitimate legal grounds for collection or use
• Ethically Questionable - Harvested from data breaches, purchased from questionable brokers, or obtained through exploitative terms of service

Putting dirty data into a clean room doesn't clean it. The data remains unconsented, inaccurate, and legally problematic. The clean room technology just obscures the original violations.

Companies know this. Privacy compliance professionals have explicitly raised this concern. Yet clean room adoption continues because the primary benefit to corporations—linking disparate customer and behavioral databases—outweighs their concern about data quality. The clean room provides technical cover for pooling data that wouldn't survive scrutiny outside the protected environment.

The Configuration Conundrum: Weak Privacy by Default

Here's where the FTC's criticism becomes specific and damaging: most data clean room implementations come with weak privacy settings by default. Companies can dial up or dial down privacy protections based on business needs.

Want stronger privacy? Increase the noise in differential privacy, reduce output granularity, and impose stricter access controls. Want better business results? Reduce privacy protections to improve data utility, increase output detail, and allow broader data access.

Most companies choose the latter. They implement clean rooms specifically to maximize business value, which means minimizing privacy constraints. The result is clean room configurations that provide minimal privacy protection despite being theoretically capable of strong protection.

This configuration flexibility is a feature the industry uses for marketing—"configure it for your privacy needs." But in practice, it's a vulnerability because most companies configure for business needs, not privacy protection.

The Re-Identification Problem: Why Your Anonymized Data Is Safer Outside the Clean Room

Here's the paradox at the heart of data clean rooms: your data is often safer remaining separated across multiple companies than it is once pooled inside a supposedly secure clean room. The clean room's primary function is linking data that was previously separate, making re-identification easier, not harder.

The Linking Attack Vector

Before data enters a clean room, your information exists in multiple separate systems:

• Your banking data is at your bank
• Your shopping behavior is with retailers
• Your mobile activity is with app publishers
• Your health information is with healthcare providers
• Your location data is with mobile networks

These separate systems create natural fragmentation that makes comprehensive re-identification difficult. A hacker who compromises your bank account data doesn't automatically get your health records or shopping history.

A data clean room collapses this fragmentation by design. It explicitly links these previously separate databases. In the process, it creates a single comprehensive profile of you that combines all these separate data streams.

Once that linking happens inside the supposedly secure clean room, the re-identification risk becomes catastrophic. Someone who gains access to just one subset of your data now has access to your complete profile. A breach of mobile advertising data suddenly exposes your financial data, health information, and shopping history because the clean room has linked everything together.

The Scaling Problem: More Data = More Vulnerability

As more companies use data clean rooms and pool data in the same environments, the re-identification risk scales exponentially. A data clean room with two participating companies linking two datasets has some vulnerability. A marketplace clean room with fifty companies contributing data from fifty different sources creates an re-identification nightmare.

The data clean room industry is moving toward marketplace models where many companies contribute data to shared environments. These marketplaces are essentially re-identification factories—the more data aggregated and linked, the easier re-identification becomes.

Companies are aware of this scaling problem. Some clean room providers are implementing output controls to prevent excessive aggregation. But these controls create a business tension: restrictions that protect privacy reduce the utility of the marketplace for data buyers.

The DisappearMe.AI Response: Protecting Yourself From the Clean Room Economy

Data clean rooms represent a new threat vector that extends beyond traditional data brokers and public records. They're sophisticated systems specifically designed to link your data across previously separate domains. Individual opt-out approaches don't work against clean rooms because they're B2B systems operating in business-to-business environments rather than business-to-consumer channels.

Protecting yourself from data clean rooms requires a multi-layered strategy that addresses both what's already known about you and what can be inferred from future data collection.

Layer 1: Minimize Source Data Across All Companies

The fundamental defense against clean room re-identification is having less data to be linked in the first place. This means:

• Removing yourself from data brokers that feed clean room systems (Acxiom, Experian, LexisNexis, Oracle, etc.)
• Minimizing the data you provide to retailers, app publishers, and service providers
• Using privacy-protective alternatives that collect less data
• Compartmentalizing identity so you have separate profiles in separate companies

Layer 2: Prevent Data Linkage Across Domains

Once data exists, preventing linkage is the second defense:

• Use different email addresses for different services so no single identifier links your profiles
• Use different phone numbers for different contexts
• Avoid using your real name on services where anonymity is possible
• Use virtual credit cards and payment methods that don't link transactions
• Prevent behavioral patterns from being too unique and identifiable

Layer 3: Monitor for Unauthorized Clean Room Participation

Companies sometimes include your data in clean rooms without explicit consent or disclosure. Protections include:

• Reading privacy policies and opt-out provisions specifically about data sharing and clean room participation
• Looking for language about "data partnerships," "collaborative analytics," or "joint analysis"
• Understanding that "anonymous" or "aggregated" outputs from clean rooms may still expose you
• Requesting that companies confirm whether your data has been enrolled in clean room partnerships

Layer 4: Understand the Legal Basis for Clean Room Processing

Most data clean room processing lacks clear legal basis:

• Companies often don't have consent for clean room use
• They may not have legitimate interest justifications that survive scrutiny
• Clean room participation may violate terms of service under which data was originally collected

Under GDPR and similar regulations, you have rights to:

• Request that companies tell you if your data is used in clean rooms
• Object to processing for clean room purposes
• Request deletion of data used in unauthorized clean rooms

Layer 5: Advocate for Regulatory Protection

Individual protections only go so far. Systemic protection requires regulatory intervention:

• Support regulatory efforts to restrict clean room usage
• Advocate for strong output controls that prevent individualized targeting from clean room results
• Push for explicit consent requirements before data can be enrolled in clean rooms
• Demand transparency about which companies are participating in which clean rooms

Frequently Asked Questions

Q: Are data clean rooms legal?

Data clean rooms themselves are legal. However, how companies use them may not be. The FTC has warned that using clean rooms to evade privacy obligations, make false privacy claims, or process data without proper legal basis is illegal. The legality depends on the specific implementation, the data being processed, and the purposes for which it's used.

Q: How can I tell if my data is in a data clean room?

You often can't. Most data clean room participation happens without explicit notification to individuals. Some companies disclose clean room participation in privacy policies under language like "data partnerships" or "collaborative analytics." You can check company privacy policies, contact companies directly, and use privacy rights under GDPR or state laws to request confirmation of clean room participation.

Q: Can I opt out of having my data in data clean rooms?

In some jurisdictions with privacy laws, you can request deletion of your data, which would remove it from clean rooms. You can also exercise privacy rights to request information about clean room participation and object to such processing. However, in many jurisdictions without comprehensive privacy laws, your options are limited. This is why DisappearMe.AI focuses on removing your data from brokers that supply clean room systems.

Q: Are the privacy technologies used in data clean rooms effective?

Differential privacy, secure multiparty computation, and encryption-in-use can provide genuine privacy protection when implemented correctly and with strong parameters. However, most commercial clean room implementations prioritize data utility over privacy, meaning they use weaker versions of these technologies. The theoretical capability exists for strong protection, but the practical implementation often provides minimal protection.

Q: How does re-identification happen with supposedly anonymized data?

Re-identification happens through linking supposedly anonymous data with external information sources, exploiting behavioral uniqueness of individuals, using demographic data combined with public records, and applying machine learning to infer sensitive attributes. The more data in a dataset, the more unique patterns exist, and the easier re-identification becomes.

Q: Should I be concerned about data clean rooms even if I've removed myself from data brokers?

Yes. If your data has been sold to data brokers that participate in clean rooms, or if companies you interact with directly participate in clean rooms, your information could still be involved. Additionally, new data brokers and clean room participants continuously enter the market, creating new exposure. Ongoing vigilance is necessary.

Q: How does DisappearMe.AI protect against data clean room exposure?

DisappearMe.AI protects against clean room exposure by: removing your data from data brokers that serve as input sources for clean rooms, monitoring for new data broker participation and clean room enrollment, implementing behavioral strategies that minimize the data you generate in the first place, and providing guidance on privacy rights you can exercise regarding clean room participation.

Q: Will clean rooms eventually become more privacy-protective?

The trajectory is uncertain. If regulators impose strong output controls and mandate explicit consent, clean rooms could become more privacy-protective. However, if industry self-regulation prevails, clean rooms will likely remain privacy theater with minimal actual protection. The outcome depends on regulatory pressure and consumer demand for privacy protection.

Q: What's the difference between a data clean room and a customer data platform (CDP)?

CDPs unify customer data from multiple internal sources into a single profile for a single company. Data clean rooms pool data from multiple companies for analysis while supposedly preventing raw data sharing. However, both systems involve comprehensive linking of individual data, and both create re-identification vulnerabilities.

About DisappearMe.AI

DisappearMe.AI recognizes that privacy threats evolve as technology advances. Data clean rooms represent a new category of threat—not just collection and sale of your data, but sophisticated linking and re-identification infrastructure that corporations are pooling behind supposedly privacy-protective technology.

The platform addresses clean room threats through comprehensive protection: removing your data from brokers that feed clean room systems, monitoring for new clean room enrollment, implementing individual privacy strategies, and providing guidance on privacy rights.

More fundamentally, DisappearMe.AI recognizes that individual protections, while essential, aren't sufficient against clean rooms operating in B2B spaces. Long-term protection requires regulatory intervention that restricts clean room functionality, mandates explicit consent for clean room participation, enforces strong output controls, and increases transparency about clean room operations.

The goal is helping individuals disappear from the data economy entirely—preventing the linking and re-identification that clean rooms enable while advocating for the regulatory changes necessary to constrain this growing threat to privacy.

References

• Decentriq. (2025). "What is a data clean room? Complete guide (2025 edition)." Retrieved from https://www.decentriq.com/article/what-is-a-data-clean-room

• DinMo. (2025). "Data Clean Rooms (DCRs): Definition, How They Work, and Uses." Retrieved from https://www.dinmo.com/modern-data-stack/mds-tools/data-clean-room/

• MarTech. (2025). "Data clean rooms: What you need to know." Retrieved from https://martech.org/data-clean-rooms-what-you-need-to-know/

• Snowflake. (2025). "What Is a Data Clean Room? How It Works and Use Cases." Retrieved from https://www.snowflake.com/en/fundamentals/what-is-a-data-clean-room/

• Plain Concepts. (2025). "What are Data Clean Rooms and why do you need them in your business?" Retrieved from https://www.plainconcepts.com/data-clean-rooms/

• Rubrik. (2025). "What is a Data Clean Room?" Retrieved from https://www.rubrik.com/insights/data-clean-room

• AdExchanger. (2024). "The 2024 Privacy Trends We'll Be Keeping An Eye On In 2025." Retrieved from https://www.adexchanger.com/data-privacy-roundup/the-2024-privacy-trends-well-be-keeping-an-eye-on-in-2025/

• Duality Tech. (2025). "What is a Data Clean Room? What DCRs Can (and Can't) Do." Retrieved from https://dualitytech.com/blog/data-clean-room/

• Didomi. (2024). "What are data clean rooms, and how do they work?" Retrieved from https://www.didomi.io/blog/data-clean-rooms

• AdMonsters. (2024). "The FTC Casts Shade on Data Clean Room Privacy Promises." Retrieved from https://www.admonsters.com/the-ftc-casts-shade-on-data-clean-room-privacy-promises/

• Decentriq. (2025). "Understanding anonymization vs. pseudonymization - Why the difference matters when choosing a data clean room." Retrieved from https://www.decentriq.com/article/anonymization-vs-pseudonymization-why-the-difference-matters-when-choosing-a-data-clean-room

• Sigma Computing. (2024). "The Hidden Risks Of Poor Data Anonymization." Retrieved from https://www.sigmacomputing.com/blog/data-anonymization

• AccelData. (2024). "Maximizing Data Security with Clean Rooms: Key Benefits & Practices." Retrieved from https://www.acceldata.io/blog/maximizing-data-security-with-clean-rooms-key-benefits-practices

• Privacy Analytics. (2024). "10 Best Practices for Re-identification Risk Determinations." Retrieved from https://privacy-analytics.com/resources/articles/10-best-practices-for-rrds

• Sustainability Directory. (2025). "How Can Data Anonymization Techniques Fail to Prevent Re-Identification and Subsequent Discrimination?" Retrieved from https://lifestyle.sustainability-directory.com/learn/how-can-data-anonymization-techniques-fail-to-prevent-re-identification-and

• Experian Marketing. (2024). "A Guide To Data Clean Rooms." Retrieved from https://www.experian.com/marketing/resources/collaboration/data-clean-rooms-guide

• NIH/PMC. (2017). "Re-identification Risks in HIPAA Safe Harbor Data: A study of United States medical data." Retrieved from https://pmc.ncbi.nlm.nih.gov/articles/PMC6344041/

• AdExchanger. (2024). "The FTC Thinks Data Clean Rooms May Have A Few Dusty Corners." Retrieved from https://www.adexchanger.com/data-privacy-roundup/the-ftc-thinks-data-clean-rooms-may-have-a-few-dusty-corners/

• IAPP. (2024). "Data clean rooms: An adtech privacy solution?" Retrieved from https://iapp.org/news/a/data-clean-rooms-an-adtech-privacy-solution