The Collapse That Changed Everything
When 23andMe filed for bankruptcy in March 2025, it held the genomic data of approximately 15 million people — the largest consumer genetics database ever assembled. The company's collapse was the result of a familiar combination of overextension, a catastrophic data breach in 2023 that compromised nearly 7 million accounts, and a business model that proved unable to convert its extraordinary data asset into sustainable revenue. What was less familiar was the governance problem it created: what happens to the genomes of millions of people when the company holding them becomes insolvent?
The answer, it turned out, was that existing legal frameworks were largely unprepared for the question. The Genetic Information Nondiscrimination Act (GINA) of 2008 — the primary federal statute governing genetic data in the United States — was designed to prevent discrimination in employment and health insurance. It said nothing about what should happen to a genomic database in bankruptcy proceedings. HIPAA, which governs health data held by covered entities, did not apply to a direct-to-consumer genetics company in the way it would apply to a hospital. The Federal Trade Commission had authority over deceptive practices but limited tools for the specific problem of genomic data in asset liquidation.
The result was a scramble. State attorneys general intervened. Consumer advocates filed objections. The bankruptcy court grappled with questions that had no clear legal precedent. By July 2026, a coalition of 42 state attorneys general had secured an $18 million settlement from 23andMe's bankruptcy estate to address the 2023 data breach. In October 2026, the California Attorney General secured a separate $6 million settlement with Chrome Holding Co. — 23andMe's successor — resolving challenges regarding data security failures and the unauthorised transfer of genetic samples during the bankruptcy proceedings. The acquiring entity, now operating as the 23andMe Research Institute, is bound by enhanced security requirements, mandatory risk analysis, and a commitment to honour consumer requests for data deletion and sample destruction.
These settlements are significant. But they are also, in a meaningful sense, retrospective — remedies for a harm that had already occurred, rather than a framework that would prevent the same harm from occurring again. The more consequential response has been legislative.
The 2026 State Legislative Wave
Throughout 2026, at least five states — South Dakota, Vermont, Rhode Island, Connecticut, and Utah — passed legislation specifically addressing the governance of direct-to-consumer genetic testing. The wave was directly catalysed by the 23andMe collapse, but it drew on a broader recognition that the existing patchwork of federal and state privacy law was inadequate for the specific characteristics of genomic data.
The dominant model in the new legislation is what might be called the express consent framework. South Dakota, Vermont, Rhode Island, and Connecticut all enacted laws requiring affirmative, specific consent — not an opt-out, but an active choice — for collecting, using, or sharing genetic data. These laws typically require separate consent for different purposes: initial testing, sample storage, and third-party data sharing are treated as distinct decisions, each requiring its own authorisation. The logic is that genomic data is not like other personal data: it is permanent, it is uniquely identifying, it reveals information about family members who have not consented to anything, and it cannot be changed if compromised.
Connecticut's legislation goes further than the others in a conceptually significant way. It explicitly grants residents a property interest in their biological samples and genetic test results — a departure from the traditional privacy law framework, which classifies genetic data as sensitive personal information subject to protection but does not treat it as property that the individual owns. The property framing has significant implications: it suggests that individuals have not merely a right to privacy in their genomic data but a right to control its use, including its commercial exploitation, in a way that resembles ownership rather than mere protection.
Utah's legislation (HB 182, effective 2028) takes a different approach, focusing on national security rather than individual rights. It prohibits the use of sequencing equipment or software linked to designated "foreign adversary" countries and bans the storage of genetic data in those jurisdictions. This reflects a growing recognition — shared by several other states that have introduced similar bills — that genomic data is not merely a privacy concern but a biosecurity concern: a national strategic resource that foreign adversaries might seek to acquire at scale.
"Genomic data is not like other personal data: it is permanent, uniquely identifying, reveals information about family members who have not consented, and cannot be changed if compromised. The express consent model reflects a belated recognition of these distinctive characteristics."
Genomic data is not like other personal data: it is permanent, uniquely identifying, reveals information about family members who have not consented, and cannot be changed if compromised. The express consent model reflects a belated recognition of these distinctive characteristics.
The European Framework: Special Category Data and the Anonymisation Problem
The European approach to genomic data governance operates within a more established framework — the GDPR's treatment of genetic data as a "special category" of personal data under Article 9, subject to heightened protection and processing restrictions. But 2026 has brought significant developments even within this more mature regulatory environment.
In April 2026, the European Data Protection Board (EDPB) adopted Guidelines 1/2026, providing substantial clarity for the research enterprise. The guidelines establish six indicative factors for determining whether data processing qualifies as "scientific research" — a designation that allows for broader processing permissions. They clarify that further processing of personal data for scientific research is presumed compatible with the original collection purpose, removing the need for a separate compatibility test in many scenarios. And they confirm that "broad consent" is permissible for scientific research within defined areas, provided the research aligns with the data subject's reasonable expectations.
More technically significant is the EDPB's draft Guidelines 02/2026 on Anonymisation, which grapple directly with the distinctive challenge that genomic data poses for de-identification. The core problem is that genetic data is inherently linked to family members: individuals may be identifiable through the genetic information of relatives, even if the primary subject's direct identifiers have been removed. This means that the concept of anonymisation — which is central to the GDPR's framework, since truly anonymous data falls outside its scope — is fundamentally more complex for genomic data than for other data types.
The draft guidelines address this through a "recipient-specific" approach to anonymisation: a dataset may be considered anonymous for a particular recipient if that recipient lacks the means to re-identify the data, even if the original controller retains a key that would allow re-identification. This is a pragmatic solution to a genuine problem, but it creates a two-tier system in which the same dataset is simultaneously anonymous (for the recipient) and personal data (for the controller) — a distinction that will require careful management in practice.
The Global Divergence: Three Models of Genomic Sovereignty
A comparative analysis published by Stanford Law School in September 2026 identifies three distinct models of genomic data governance that are now operating in parallel globally, with limited convergence between them.
The European model treats genetic data as a fundamental rights issue. The GDPR's substantive, identity-based definition — genetic data is protected due to its inherent nature, regardless of who holds it or in what context — reflects a constitutional commitment to human dignity and bodily integrity. The framework is comprehensive but complex, and its application to research contexts has required ongoing regulatory guidance to prevent it from becoming an obstacle to legitimate scientific work.
The American model remains fragmented. Federal law — GINA, HIPAA, the FTC Act — provides partial coverage that depends heavily on the identity of the entity holding the data and the context in which it was collected. The 2026 state legislative wave has added new layers of protection in specific jurisdictions, but the result is a patchwork that creates compliance complexity for national and international actors and leaves significant gaps in coverage for individuals in states that have not acted.
The Chinese model treats genomic data as a national strategic resource. Under rules updated in November 2026, China mandates that activities involving the collection, preservation, and use of human genetic resources — including derived genetic information — must not endanger national security, public health, or the public interest. Foreign organisations seeking to utilise Chinese genetic resources for scientific research are required to partner with Chinese institutions and obtain approval from the National Health Commission. This paradigm treats genomic data as an intersection of data protection and state sovereignty, distinct from the EU's focus on individual fundamental rights.
These three models are not merely different regulatory approaches — they reflect fundamentally different conceptions of what genomic data is and who has legitimate claims over it. The divergence creates significant challenges for international research collaborations, for multinational companies operating across jurisdictions, and for the development of global genomic databases that could accelerate medical research.
"Three models of genomic sovereignty are now operating in parallel globally: the European fundamental rights framework, the American patchwork, and the Chinese strategic resource model. Their divergence is not merely regulatory — it reflects fundamentally different conceptions of what genomic data is and who has legitimate claims over it."
Three models of genomic sovereignty are now operating in parallel globally: the European fundamental rights framework, the American patchwork, and the Chinese strategic resource model. Their divergence is not merely regulatory — it reflects fundamentally different conceptions of what genomic data is and who has legitimate claims over it.
Indigenous Data Sovereignty: The Collective Dimension
The governance frameworks described above share a common assumption: that the relevant unit of genomic data governance is the individual. But for Indigenous communities, this assumption is itself a form of epistemic colonialism. Genomic data about an individual member of an Indigenous community is also, in a meaningful sense, data about the community — its ancestry, its health vulnerabilities, its genetic heritage. Individual consent frameworks, however robust, do not capture this collective dimension.
Australia's National Health Genomics Policy Framework and Implementation Plan 2026–2030, published in September 2026, represents one of the most significant institutional responses to this challenge. The framework mandates Aboriginal and Torres Strait Islander-led governance of genomic data, ensuring alignment with Indigenous Data Sovereignty principles. This goes beyond requiring community consultation: it requires that Indigenous communities have genuine decision-making authority over research involving their genetic resources, including the right to refuse participation, to set conditions on data use, and to benefit from research outcomes.
The academic literature on Indigenous genomic data sovereignty, including a March 2026 paper in the European Journal of Human Genetics, emphasises that researchers must move beyond simple informed consent toward collaborative, community-led models that respect cultural protocols and address the legacy of colonial research practices. The history of genomic research involving Indigenous communities includes numerous cases in which data collected for one purpose was used for another without community knowledge or consent — a pattern that has generated justified mistrust and that individual consent frameworks alone cannot address.
The Indigenous data sovereignty movement is also raising questions about benefit-sharing that go beyond the immediate research context. If genomic data from Indigenous communities contributes to the development of pharmaceutical products or diagnostic tools, what obligations do the beneficiaries of that research have to the communities whose data made it possible? These questions do not yet have settled legal answers in most jurisdictions, but they are increasingly being raised in research ethics frameworks and in negotiations between research institutions and Indigenous communities.
The Anonymisation Illusion and the Re-identification Risk
One of the most significant technical developments in genomic data governance in 2026 has been the growing recognition that genomic data cannot be reliably anonymised using conventional techniques. Research has consistently demonstrated that individuals can be re-identified from genomic datasets even when direct identifiers have been removed, using a combination of the genetic data itself, publicly available genealogical databases, and demographic information.
The implications for existing governance frameworks are significant. Many research data-sharing arrangements, and many of the consent frameworks that underpin them, are premised on the assumption that de-identified genomic data is not personal data and therefore not subject to the full rigour of data protection law. If that assumption is incorrect — if genomic data is effectively always identifiable — then the entire architecture of genomic research data governance needs to be reconsidered.
The EDPB's draft anonymisation guidelines acknowledge this problem but do not fully resolve it. The recipient-specific approach to anonymisation provides a workable framework for specific data-sharing arrangements, but it does not address the broader question of what happens when the technical means of re-identification become more widely available — as they inevitably will, as genealogical databases grow and as AI-powered re-identification techniques improve.
The Property Rights Frontier
Connecticut's decision to frame genomic data governance in terms of property rights rather than privacy rights may prove to be the most consequential legislative innovation of 2026. The property framing has several significant implications that the privacy framing does not.
If genomic data is property, then using it without compensation may constitute a form of taking, not merely a privacy violation — a framing with significant implications for the business models of genomic research companies and health data brokers.
First, it suggests that individuals have not merely a right to prevent others from misusing their genomic data, but a right to benefit from its use — including its commercial exploitation. If genomic data is property, then using it without compensation may constitute a form of taking, not merely a privacy violation. This has significant implications for the business models of genomic research companies, pharmaceutical companies, and health data brokers.
Second, the property framing provides a clearer basis for addressing the bankruptcy scenario that the 23andMe collapse exposed. If genomic data is property belonging to the individuals who provided it, then it cannot simply be transferred to a new owner as part of a bankruptcy estate — or if it can, the individuals whose property it is have claims that must be addressed in the bankruptcy proceedings.
Third, the property framing creates a basis for collective governance of genomic data that the privacy framing does not easily support. If a community's genomic data is collective property, then decisions about its use require collective consent — a framework that aligns more naturally with Indigenous data sovereignty principles than the individual consent model that dominates current privacy law.
Whether the property framing will spread beyond Connecticut, and whether it will survive legal challenges from the genomic research and commercial genetics industries, remains to be seen. But it represents a genuinely novel approach to a problem that existing frameworks have struggled to address.
What Adequate Governance Would Require
The 23andMe collapse, the 2026 legislative wave, the EDPB's evolving guidance, and the Indigenous data sovereignty movement all point toward a set of requirements that adequate genomic data governance would need to satisfy.
It would need to treat genomic data as categorically different from other personal data — not merely as sensitive information requiring heightened protection, but as a distinctive type of data whose permanence, heritability, and re-identification risk require a governance framework specifically designed for its characteristics.
It would need to address the collective dimension of genomic data — the fact that an individual's genome contains information about their family members and, in some cases, their community — in a way that individual consent frameworks cannot.
It would need to address the insolvency scenario — establishing clear rules for what happens to genomic databases when the companies holding them fail, including whether genomic data can be transferred as a commercial asset and what rights individuals retain in that scenario.
And it would need to address the international dimension — the fact that genomic data flows across borders, that research collaborations are inherently international, and that the three divergent governance models currently operating globally create both compliance complexity and opportunities for regulatory arbitrage.
None of these requirements is fully met by any existing framework. The 2026 legislative wave represents genuine progress, but it is state-level, fragmented, and reactive. The EDPB's guidelines represent genuine sophistication, but they operate within a framework that was not designed for genomic data's distinctive characteristics. The Indigenous data sovereignty movement represents a genuinely different conception of what genomic governance should look like, but it has not yet been integrated into mainstream regulatory frameworks.
The genome is contested territory. The governance frameworks adequate to that contest are still being built.



