Study of depression management and suicide prevention apps shows many aren't following best practices and some even contain inaccurate hotline contact info
Nicole Wetsman / The Verge : Tweets: @coyneoftherealm and @ethics13 Tweets: James C.Coyne / @coyneoftherealm : Apps don't provide reliable help for #suicideprevention: A study found nonfunctioning crisis hotline numbers in six apps: https://www.theverge.com/... @ethics13 : Nothing does, except.... Being in a life of someone you know intimately, seeing or hearing the signs, and working with that someone to guide them out. The above takes a lot of energy, time, and investment. Most people do not have the stamina. So we have high suicides. https://twitter.com/...
Context & Ripple Effects
This study lands in the middle of a widening scrutiny cycle for consumer mental health tech. Experts had already questioned Facebook's opaque approach to flagging possible suicide threats to police the year before, and weeks after this piece ran, reporting surfaced on how therapy apps like Better Help notify third parties when opened and when a user reports suicidal thoughts (Better Help's data-sharing practices).
What makes the finding consequential is placement: these apps sit on the crisis pathway, so a dead hotline number is not a bug but a failed referral. The field's subsequent arc — a research project using smartphone and Fitbit data to detect high-danger periods (AI analysis of at-risk people's wearable data), and expert criticism of Koko for routing at-risk users on Facebook, Tumblr, and Discord to a chatbot (Koko's chatbot study) — reads as a search for alternatives to exactly the unvalidated tools this study audited.
First-order effects
- Users who open any of the six apps containing nonfunctioning crisis hotline numbers may reach no one at the moment of need, and those developers face immediate pressure to verify their contact information.
- James Coyne and his co-authors hand clinicians, reviewers, and journalists a concrete audit standard — adherence to published best practices — that every depression-management app now has to answer against.
Second-order effects
- Platforms that route at-risk users into third-party tools inherit the quality problem: Facebook's flagging approach was already contested, and Koko's later chatbot experiment drew expert criticism for pointing vulnerable users at an unproven intervention.
- Therapy apps like Better Help, already reported to signal third parties on app opens and suicidal-thought disclosures, now carry compounded quality-plus-privacy liability that raises trust costs across the whole category.
Third-order effects
- If static self-help apps keep failing basic audits, research funding and talent migrate toward passive sensing models like the smartphone-and-wearables detection project, leaving app stores distributing mental health tools with no validation gate in place.
- Crisis support consolidates around verified channels — live hotlines and clinical contact — while unvalidated apps get pushed out of the safety pathway entirely, a split regulators have yet to formalize.
The trend: Consumer mental health software is being pushed from unvalidated self-help apps toward clinically audited tools and passive sensor-based detection, with each quality scandal tightening the bar for what counts as a legitimate crisis resource.