Several life insurance firms testing tech that estimates life expectancy with analysis of selfie pictures
Barbara Marquand / USA Today :
Context & Ripple Effects
The selfie-underwriting trials land mid-arc in an industry steadily widening what counts as underwriting data. A year later, John Hancock would go further than any rival, announcing it will only sell policies that track fitness and health data via wearables and smartphones. ProPublica's investigation that summer had already shown health insurers working with data brokers to collect lifestyle details on hundreds of millions of Americans.
What the selfie tests add is a new input class: a facial image volunteered in seconds, run through models that infer mortality risk without an exam or a blood draw. It is the same logic Bloomberg later profiled in [[a:879863|Death Clock, an app predicting a user's date of death from an AI model trained on 1,200+ life expectancy studies]] — mortality prediction moving from clinics to cameras.
First-order effects
- Applicants to the testing insurers can be priced on a facial analysis they may not realize feeds an actuarial model, replacing or shortcutting traditional medical-exam steps for some policies.
- The insurers running these pilots gain a faster, cheaper risk-assessment channel at the point of quote, ahead of rivals still dependent on exams and self-reported health.
Second-order effects
- Competitors face pressure to match the friction-free application or concede the quick-issue segment, accelerating the shift New York Times coverage documented toward premiums adjusted by new forms of surveillance across life, car, and home insurance.
- Data brokers gain a fresh market: facial images and derived biomarkers become another lifestyle dataset to package for insurers, extending the pipeline ProPublica exposed.
Third-order effects
- If facial inference proves predictive enough to price on, underwriting migrates from episodic exams to passive biometric estimation — with no regulatory framework yet governing it, echoing STAT's finding that insurers already deploy unregulated predictive algorithms in coverage decisions.
- Consumer consent becomes the structural battleground: as with John Hancock's tracking-only policies, the choice narrows to sharing biometric data or paying more for coverage, inviting scrutiny of whether such pricing discriminates by appearance, age, or ethnicity.
The trend: Life insurance pricing is shifting from medical exams and disclosed health history toward passive AI inference drawn from everyday consumer data — selfies today, wearables and lifestyle streams around them.