Tech companies should automatically turn off irrelevant ads based on info users share, such as ads for baby gear when a pregnant person has a stillborn child
Dear Tech Companies: — I know you knew I was pregnant. It's my fault, I just couldn't resist those Instagram hashtags — #30weekspregnant, #babybump.
Context & Ripple Effects
This opinion piece lands mid-arc in a running argument about how platforms handle inferred health status. It follows reporting that fitness and health trackers like Apple Health and Watch ignore pregnancy entirely — the same ecosystem that happily harvests #30weekspregnant hashtags for targeting but drops the user the moment the pregnancy ends badly.
The plea also predates a darker turn: by 2022, Reuters reported that [[a:980210|law enforcement could compel tech companies to hand over pregnancy-related search histories and location data]] under state abortion restrictions, turning what this author calls a tasteless ad problem into a legal exposure problem.
First-order effects
- Users who share pregnancy information on platforms like Instagram remain locked into baby-gear and nursery targeting after pregnancy loss, because ad systems have no mechanism to un-learn an inference they were built to exploit.
- The named platforms — Instagram and its ad peers — absorb direct reputational backlash each time a grieving user documents the mismatch between what they shared and what gets served.
Second-order effects
- Backlash-driven policy fixes become the pattern: Google's requirement that abortion-services advertisers disclose whether they provide abortions shows platforms responding to exactly this kind of pressure with advertiser-side rules rather than changes to inference itself.
- Advertisers gain a new brand-safety calculus around sensitive-life-event targeting, since serving grief-mismatched ads damages the brand paying for the impression as much as the platform delivering it.
Third-order effects
- If the pattern holds, pregnancy and other health inferences migrate from pure monetization assets to governed liabilities — the same data that powers targeting can be subpoenaed, as the 2022 state-law reporting suggests, pushing platforms toward deliberate data minimization on sensitive categories.
- The structural endpoint is a split ad market: platforms that can prove they govern sensitive inferences compete for health-adjacent advertising, while those that cannot retreat from the category altogether — a shift already visible in Apple's slow fix of abortion searches routing to adoption and fertility centers.
The trend: Health inferences harvested for ad targeting are being reclassified by backlash and law alike from monetization assets to liabilities requiring explicit platform governance.