Loading
What changed between revision 1 and revision 2, compared over the article's markdown source with its frontmatter — so a change to status, evidence or reversibility shows up as the edit it is.
Older · revision 1
Import· initial corpus import, not an edit
Initial import of content/genetic-discrimination.md — the filesystem corpus, unchanged. Not an edit.
12.3 kB
Revision 1
Revision 2
lines 7–13 → 7–136 unchanged lines not shown
categories: ["society", "genetics"]categories: ["society", "genetics"]tags: ["law", "privacy", "insurance", "employment", "gina", "genomics"]tags: ["law", "privacy", "insurance", "employment", "gina", "genomics"]summary: "Differential treatment of people on the basis of genetic information, and the uneven legal regime that restricts it in employment and health cover but rarely elsewhere."summary: "Differential treatment of people on the basis of genetic information, and the uneven legal regime that restricts it in employment and health cover but rarely elsewhere."updated: "2026-07-27"updated: "2026-10-29"issues: ["The consumer-genomics breach, the 2025 bankruptcy, and the Florida and Australia changes are uncited."]issues: ["The consumer-genomics breach, the 2025 bankruptcy, and the Florida and Australia changes are uncited."]------ lines 25–92 → 25–9211 unchanged lines not shown
}}`````` **Genetic discrimination** is the use of a person's genetic information — a test result, a family history, or a predicted risk — to treat them less favourably in insurance, employment, education, or another domain where the information is not relevant to present ability. The concern predates the ability to sequence a genome cheaply and has produced a patchwork of protections that is strongest where the risk was anticipated and weakest where the information is now actually used.**Genetic discrimination** is the use of a person's genetic information, such as a test result, a family history, or a predicted risk, to treat them less favourably in insurance, employment, education, or another domain where the information is not relevant to present ability. The concern predates cheap genome sequencing. It has produced a patchwork of protections that is strongest where the risk was anticipated and weakest where the information is now actually used. ## What is being protected## What is being protected The canonical case is a person who learns they carry a pathogenic variant for a late-onset condition — Huntington's disease is the standard example because penetrance is near total and onset is decades away — and is then refused cover or employment on the strength of a prediction about a future they have not yet reached. The harm has two parts. One is the adverse decision. The other is the chilling effect: people who fear the consequences decline testing that would benefit them clinically, which is the effect surveys of at-risk families most consistently find. The same reluctance could extend to predictive tests that are not genetic at all, including the direct-to-consumer panels covered in [[consumer-blood-testing]] and the multi-cancer screening assays described in [[nanoparticle-diagnostics]], where a positive result may reach an underwriter long before it becomes a diagnosis.The canonical case is a person who learns they carry a pathogenic variant for a late-onset condition and is then refused cover or employment on the strength of a prediction about a future they have not yet reached. Huntington's disease is the standard example, because penetrance is near total and onset is decades away. The harm has two parts. One is the adverse decision. The other is the chilling effect, in which people who fear the consequences decline testing that would benefit them clinically; surveys of at-risk families find this effect most consistently. The same reluctance could extend to predictive tests that are not genetic at all, including the direct-to-consumer panels covered in [[consumer-blood-testing]] and the multi-cancer screening assays described in [[nanoparticle-diagnostics]], where a positive result may reach an underwriter long before it becomes a diagnosis. Whether genetic information deserves special treatment is contested. The position called genetic exceptionalism holds that it is distinctively sensitive, because it is predictive, shared with relatives, and immutable — the last of which ceases to be true if the editing methods described in [[somatic-gene-therapy]] mature.[^murray1997] Critics note that family history, cholesterol, and an HIV test are also predictive and shared, that most genetic variants carry weak probabilistic information rather than a verdict, and that carving out genetics produces an incoherent privacy regime in which the same fact is protected or unprotected depending on how it was learned.Whether genetic information deserves special treatment is contested. The position called genetic exceptionalism holds that it is distinctively sensitive, because it is predictive, shared with relatives, and immutable. The last of these ceases to be true if the editing methods described in [[somatic-gene-therapy]] mature.[^murray1997] Critics note that family history, cholesterol, and an HIV test are also predictive and shared, that most genetic variants carry weak probabilistic information rather than a verdict, and that carving out genetics produces an incoherent privacy regime in which the same fact is protected or unprotected depending on how it was learned. Sequencing has strengthened the critics' case. Everyone carries variants associated with elevated risk of something. Once whole-genome data is routine, protecting "genetic test results" as a category protects almost nothing, because the discriminating input is the interpretation rather than the sequence — a point that also complicates [[polygenic-embryo-screening]], where the output is a probability rather than a diagnosis.Sequencing has strengthened the critics' case. Everyone carries variants associated with elevated risk of something. Once whole-genome data is routine, protecting "genetic test results" as a category protects almost nothing, because the discriminating input is the interpretation rather than the sequence. The same point complicates [[polygenic-embryo-screening]], where the output is a probability rather than a diagnosis. ## GINA and its gaps## GINA and its gaps The US Genetic Information Nondiscrimination Act, enacted in 2008 after more than a decade in Congress, prohibits health insurers from using genetic information in eligibility, premiums, or underwriting, and prohibits employers with fifteen or more employees from using it in hiring, firing, or terms of employment. It also bars employers from requesting or purchasing genetic information, including family medical history, with narrow exceptions.The US Genetic Information Nondiscrimination Act, enacted in 2008 after more than a decade in Congress, prohibits health insurers from using genetic information in eligibility, premiums, or underwriting, and prohibits employers with fifteen or more employees from using it in hiring, firing, or terms of employment. It also bars employers from requesting or purchasing genetic information, including family medical history, with narrow exceptions. Its exclusions define the current problem.The current problem lies in its exclusions. **Life, disability, and long-term care insurance are not covered.** These are the products for which a predictive genetic result is most actuarially useful, and in most US states insurers may lawfully ask for and act on test results. Florida became the first state to close the gap in 2020, barring life, disability, and long-term-care insurers from using genetic test information; a small number of other states have since restricted it in narrower ways.**Life, disability, and long-term care insurance are not covered.** These are the products for which a predictive genetic result is most actuarially useful, and in most US states insurers may lawfully ask for and act on test results. Florida became the first state to close the gap in 2020, barring life, disability, and long-term-care insurers from using genetic test information; a small number of other states have since restricted it in narrower ways. **Manifest disease is not covered.** Once a condition has appeared, it is a medical condition rather than genetic information, and GINA no longer applies. Protection is therefore strongest for the asymptomatic and disappears when the risk materialises.**Manifest disease is not covered.** Once a condition has appeared, it is a medical condition rather than genetic information, and GINA no longer applies. Protection is therefore strongest for the asymptomatic and disappears when the risk materialises. **Small employers, the military, and several federal programmes fall outside.** The armed services are not bound by GINA, and genetic findings can affect assignment and retention. The exclusion matters more than its size suggests, since military medicine is where several enhancement-adjacent technologies are first adopted, as [[enhancement-arms-race]] describes.**Small employers, the military, and several federal programmes fall outside.** The armed services are not bound by GINA, and genetic findings can affect assignment and retention. The exclusion matters more than its size suggests, because military medicine is where several enhancement-adjacent technologies are first adopted, as [[enhancement-arms-race]] describes. **Selection is untouched.** GINA governs what may be done with a living person's genetic information. It says nothing about screening embryos, so the practices covered in [[embryo-selection]] fall entirely outside it.**Selection is untouched.** GINA governs what may be done with a living person's genetic information. It says nothing about screening embryos, so the practices covered in [[embryo-selection]] fall entirely outside it. Enforcement has been thin. Regulators have brought a modest number of actions, most concerning employers collecting family medical history in pre-employment physicals or wellness programmes rather than sequencing anyone. The best-known US episode predates the statute: a railway company was found in 2001 to have secretly tested employees who had filed carpal-tunnel claims for a variant it believed predisposed them to the condition, and settled with the federal employment regulator the following year.Enforcement has been thin. Regulators have brought a modest number of actions, most concerning employers collecting family medical history in pre-employment physicals or wellness programmes rather than sequencing anyone. The best-known US episode predates the statute: a railway company was found in 2001 to have secretly tested employees who had filed carpal-tunnel claims for a variant it believed predisposed them to the condition, and settled with the federal employment regulator the following year. > [!caution] Absence of complaints is weak evidence> [!caution] Absence of complaints is weak evidence> Low case volume is often cited as showing that genetic discrimination is rare. It is equally consistent with the discrimination being undetectable: an applicant declined cover is rarely told which item in the file was decisive, and an underwriting model that weights a family history does not announce itself.> Low case volume is often cited as showing that genetic discrimination is rare. It is equally consistent with discrimination that cannot be detected: an applicant declined cover is rarely told which item in the file was decisive, and an underwriting model that weights a family history does not announce itself. ## Outside the United States## Outside the United States Canada enacted the strongest instrument, a 2017 statute making it a criminal offence to require a genetic test or the disclosure of results as a condition of a contract or service. Its constitutionality was challenged on federalism grounds — insurance is a provincial matter — and the Supreme Court of Canada upheld it as a valid exercise of the criminal law power in 2020.[^scc2020]Canada enacted the strongest instrument, a 2017 statute making it a criminal offence to require a genetic test or the disclosure of results as a condition of a contract or service. Its constitutionality was challenged on federalism grounds, since insurance is a provincial matter, and the Supreme Court of Canada upheld it as a valid exercise of the criminal law power in 2020.[^scc2020] The United Kingdom relies on industry self-regulation. Under a code agreed between government and insurers, companies do not ask for predictive genetic test results, with a single exception for Huntington's disease on life cover above a high threshold. It has held for two decades and has no statutory force; the equivalent Australian arrangement was a voluntary industry moratorium, and in 2024 the Australian government announced it would legislate a ban on the use of genetic test results in life insurance underwriting.The United Kingdom relies on industry self-regulation. Under a code agreed between government and insurers, companies do not ask for predictive genetic test results, with a single exception for Huntington's disease on life cover above a high threshold. The code has held for two decades but has no statutory force. The equivalent Australian arrangement was a voluntary industry moratorium, and in 2024 the Australian government announced it would legislate a ban on the use of genetic test results in life insurance underwriting. The Council of Europe's Oviedo Convention prohibits discrimination on grounds of genetic heritage among its parties, and EU data protection law treats genetic data as a special category requiring an explicit lawful basis. Neither settles the underwriting question directly, and jurisdictions have reached opposite conclusions about whether an insurer may use information the applicant already holds. The same convention supplies the only binding international prohibition on heritable modification, which is why it recurs throughout [[governance-of-genome-editing]] and the debates over [[germline-editing]].The Council of Europe's Oviedo Convention prohibits discrimination on grounds of genetic heritage among its parties, and EU data protection law treats genetic data as a special category requiring an explicit lawful basis. Neither settles the underwriting question directly, and jurisdictions have reached opposite conclusions about whether an insurer may use information the applicant already holds. The same convention contains the only binding international prohibition on heritable modification, which is why it recurs throughout [[governance-of-genome-editing]] and the debates over [[germline-editing]]. ## The adverse-selection argument## The adverse-selection argument Insurers make a real argument against prohibition. If applicants may know their genetic risk while insurers may not, the high-risk buy more cover, premiums rise, and the low-risk exit — the classic adverse-selection spiral. Where health cover is universal and publicly funded the argument does not arise, which is why the issue is sharpest in countries with private health and life markets.Insurers make a real argument against prohibition. If applicants may know their genetic risk while insurers may not, the high-risk buy more cover, premiums rise, and the low-risk exit. This is the classic adverse-selection spiral. Where health cover is universal and publicly funded the argument does not arise, which is why the issue is sharpest in countries with private health and life markets. The empirical picture is less alarming than the theory. Studies of markets operating under a genetic-testing moratorium have not found the predicted destabilisation, plausibly because the number of people holding a strongly predictive result remains small relative to the pool and because most policies are bought before anyone tests. The distributive stakes also run the other way from those in [[access-and-inequality]]: here the worry is not that a treatment is unaffordable but that knowing one may need it makes cover unobtainable. Whether that holds as polygenic risk scoring becomes routine is unknown, and it is the version of the question that matters for the next decade.The empirical picture is less alarming than the theory. Studies of markets operating under a genetic-testing moratorium have not found the predicted destabilisation, plausibly because the number of people holding a strongly predictive result remains small relative to the pool and because most policies are bought before anyone tests. The distributive stakes also run the other way from those in [[access-and-inequality]]: the worry here is that knowing one may need a treatment makes cover unobtainable, rather than that the treatment is unaffordable. Whether this holds as polygenic risk scoring becomes routine is unknown, and that version of the question is the one that matters for the next decade. ## Consumer genomics and the data itself## Consumer genomics and the data itself Direct-to-consumer testing put tens of millions of genotypes into private databases governed by terms of service rather than by health-privacy law. Three developments have made that arrangement look fragile.Direct-to-consumer testing put tens of millions of genotypes into private databases governed by terms of service rather than by health-privacy law. Three developments have made that arrangement look fragile. Law enforcement access came first: the 2018 identification of a serial offender in California through a genealogy database showed that a relative's upload can expose a person who never tested.[^erlich2018] Forensic genetic genealogy is now routine in US investigations, and the individuals whose data enables it did not consent to that use.Law enforcement access came first. The 2018 identification of a serial offender in California through a genealogy database showed that a relative's upload can expose a person who never tested.[^erlich2018] Forensic genetic genealogy is now routine in US investigations, and the individuals whose data enables it did not consent to that use. Security followed. A large consumer testing company disclosed in 2023 that attackers using credentials stolen elsewhere had accessed millions of profiles through a relative-matching feature — an attack the architecture invited, since the product's purpose is to connect one account's data to another's.Security followed. A large consumer testing company disclosed in 2023 that attackers using credentials stolen elsewhere had accessed millions of profiles through a relative-matching feature. The architecture invited the attack, since the product's purpose is to connect one account's data to another's. Insolvency completed the set. The same company entered Chapter 11 bankruptcy protection in 2025, and its genetic database was treated as an asset in a court-supervised sale. Several US state attorneys general advised residents to delete their accounts. No US federal law establishes what happens to a genomic database when its custodian fails, and the terms of service under which the data was collected are contracts that a bankruptcy court may modify.Insolvency was the third. The same company entered Chapter 11 bankruptcy protection in 2025, and its genetic database was treated as an asset in a court-supervised sale. Several US state attorneys general advised residents to delete their accounts. No US federal law establishes what happens to a genomic database when its custodian fails, and the terms of service under which the data was collected are contracts that a bankruptcy court may modify. ## Where it is heading## Where it is heading Three shifts are already straining the existing framework.Three shifts already strain the existing framework. Polygenic scores turn discrimination from a question about rare variants into a question about everyone, since a score can be computed for any trait from a standard genotype array — the mechanism behind [[genetic-enhancement-of-intelligence]] and the screening applications marketed under the heading of [[designer-babies]]. Insurers do not need a diagnosis to price a risk; a distribution will do.Polygenic scores turn discrimination from a question about rare variants into a question about everyone, since a score can be computed for any trait from a standard genotype array. The same mechanism lies behind [[genetic-enhancement-of-intelligence]] and the screening applications marketed under the heading of [[designer-babies]]. Insurers do not need a diagnosis to price a risk; a distribution is enough. Population genomic programmes are placing sequence data into national health systems, where the governance question is not whether an insurer may see it but whether a state may.Population genomic programmes are placing sequence data into national health systems, where the governance question shifts from whether an insurer may see it to whether a state may. And the arrival of heritable modification, if it ever occurs, creates a category the statutes do not contemplate: discrimination not on the basis of what a person's genome is but on the basis of what was done to it. Nothing in existing law reaches it, and the parallel debate over [[neurorights]] has begun to raise the same structural question for neural data. Whether the correct response is another category-specific statute, or a general prohibition on unjustified prediction-based discrimination of the kind [[disability-rights-and-enhancement]] scholarship has long argued for, remains unsettled in every jurisdiction that has legislated so far.Heritable modification, if it ever arrives, creates a category the statutes do not contemplate: discrimination based on what was done to a person's genome rather than on what that genome is. Nothing in existing law reaches it, and the parallel debate over [[neurorights]] has begun to raise the same structural question for neural data. Whether the correct response is another category-specific statute, or a general prohibition on unjustified prediction-based discrimination of the kind [[disability-rights-and-enhancement]] scholarship has long argued for, remains unsettled in every jurisdiction that has legislated so far. ## See also## See also removed, struck through added, underlinedLine numbers count the serialised markdown of each revision, frontmatter included.
87 lines are the same in both revisions, and 32 of them are not shown. Neither revision here is necessarily the current article — read Genetic discrimination for the text this wiki stands behind today.