Note · privacy · policy
When Period Data Becomes Legal Evidence
Information management is the control of information across its lifecycle: collection, storage, access, sharing, and deletion. Stated that way it sounds like a filing problem. It is closer to a question about power, because each stage in that lifecycle decides who can later compel the record and for what purpose. Where the lifecycle goes unregulated, the result is not only an erosion of privacy but an expansion of what the state can find out about people who never volunteered anything to it.
Menstrual tracking apps are the clearest current example. Flo and Premom were both the subject of Federal Trade Commission action over sharing users’ period data with third parties without explicit permission. That alone would be an ordinary story about consumer software behaving badly. Plenty of apps retain and sell more personally identifiable information than they need. What makes menstrual data different is what happened to the legal environment around it after Roe v. Wade was overturned. In a country where some states criminalize conduct that others protect, a record of a cycle that stopped and then resumed is not merely a health note. It can become evidence about a person, produced without their knowledge, with legal consequences that are hard to reverse.
The instinctive response is that health data is regulated. It is, but not this health data. HIPAA governs covered entities: providers, plans, clearinghouses, and their business associates. A consumer app that a person downloads and types into is generally none of those. The data is medically sensitive in substance and unprotected in law, which is the worst combination available. There is no rule requiring the app to resist a subpoena, and no rule preventing it from having built a retention policy that makes complying easy.
Once data is collected and lands in a central database, sharing can be monetized in ways that have nothing to do with the interests of the people who generated it. The deeper problem is that the data does not stay in the context it was created for. Somebody enters a symptom to predict a cycle, and the record can later be read in a courtroom to establish a timeline. Nothing in the system marks the original purpose as binding. Consent given at signup, to an app, for a feature, does not travel with the record and does not constrain the third party who receives it or the authority who later demands it.
Three trends make this worse, and they compound rather than merely add. More detailed data means the record can support more specific inferences. Longer retention means the window in which it can be demanded stays open, often long after the user has stopped thinking about the app at all. More centralization means there is a single place to send the demand. Each trend is defensible in isolation, and each is usually justified by better predictions, product continuity, or engineering simplicity. Together they produce information that is both more incriminating and easier to reach.
It is fair to note that these apps have not, as far as is publicly known, been directly subpoenaed by state or local governments for this purpose. That is not much comfort, and it is not an argument that the risk is theoretical.
The risk is credible for three reasons that hold right now. There is no legal protection preventing such data from being used this way. The apps have a demonstrated history of sharing information inappropriately, which is evidence about how they behave under pressure that costs them nothing. And resisting a state-level subpoena is a political act with real costs, which a consumer software company has little structural reason to absorb on a user’s behalf. As polarization increases and state and federal law continue to diverge, the probability rises rather than falls. Designing on the assumption that nobody will ever ask is not a privacy practice. It is a bet.
The situation is not fixed, and the useful thing about naming it as an information management failure is that information management has known remedies.
Washington’s My Health My Data Act, passed in 2023, extends HIPAA-like protection to online consumer health data, closing the definitional gap that leaves these apps outside the health privacy regime in the first place. Some European apps, Clue among them, operate under stricter rules that prevent the sharing of non-anonymous data at all. Both approaches attack the lifecycle rather than the intention. They do not ask companies to promise better; they change what is collectable, retainable, and shareable, which is the only part of the promise that survives a subpoena.
That is the point worth holding onto. The protective work is done upstream, in what the system is built to keep, not downstream in what anyone intends to do with it. A record that was never retained cannot be produced. Everything else is a policy that can be changed by whoever owns the database next.