The Black Box Project: Why the Warning Signs Were Hidden

“They seemed OK.” That is the sentence so many survivors say at some point in our support group meetings, usually the same week they walk in for the first time. Just OK. Full stop, as if the possibility had already been checked and closed.

In August 2025, a nonprofit called Stop Soldier Suicide published research that finally seeks to explain why that sentence is so often true and so completely misleading at the same time. The Black Box Project analyzed the phones of more than a hundred people who died by suicide and found that many of them kept their outward messages, calls, and conversations remarkably steady right up until the end.

As a facilitator who has sat with suicide loss survivors at SOS Madison for more than fifteen years, I have heard “they seemed OK” more times than I can count, almost always followed by the same quieter, harder question. So what did I miss. This research finally gives us something we have needed for a long time, actual evidence rather than only theory, that the missing was never a failure of attention.

If you are in crisis right now, please call or text 988 for the 988 Suicide and Crisis Lifeline. If you are a veteran or service member, you can reach the Veterans Crisis Line the same way, by calling 988 and pressing 1. Survivors of suicide loss carry an elevated risk of their own, and reaching out is never a small thing.

In This Post

• What the Black Box Project Studied
• The Facade That Fooled Everyone
• The Ten-Week Shift in Sleep
• Still Fighting: Cycles of Searching
• Money, Isolation, and Quiet Signals
• Never Only a Veteran’s Story
• What This Means for Your Guilt
• Where Prevention Goes From Here
• Questions Survivors Ask

What the Black Box Project Actually Studied

The Black Box Project takes its name from the flight data recorder, built after aviation’s worst years so investigators could learn from every incident. Stop Soldier Suicide, a nonprofit focused on veteran and service member suicide, asked a version of the same question. What if the phone in someone’s pocket held a record of the hours and weeks before they died, the way a black box holds a flight’s final moments.

Families who had lost someone to suicide answered that question by doing something genuinely brave. They loaned their person’s phone, tablet, or computer to the project’s forensic team, which handles it using the same procedures as federal law enforcement and returns it within about ninety days, hardware and data intact. Nothing about a device is shared back out in an identifiable way. The research team pools and depersonalizes everything, and the patterns that emerge only mean anything at the population level, never as a verdict on one specific phone.

Almost everything we have known about people who died by suicide has come from other people’s accounts of them, interviews conducted after the fact, or from suicide notes, words chosen with a reader in mind.

This research offers something different, a way to look inside a person’s actual thoughts and actions as they struggled through crisis, not a final note left for someone else, but the private searches and unsent drafts nobody was ever meant to see. That is a rare vantage point into not just a person’s last moment, but the thinking that led there.

These are not phones seized by the police for an investigation. They are phones a grieving parent, spouse, or sibling chose to hand over, because they wanted their person’s story to do something beyond ending in grief, a real act of generosity and, for many families, a form of honoring someone they cared for deeply. The white paper itself opens with a promise to those families, that their trust would be treated as sacred ground, not just as evidence. And to be fair, the paper is quite a bit technical, which is why I chose to write a simpler version about it as a blog post.

By the time this white paper was published, the team had analyzed data from more than a hundred donated devices, still a modest sample, and several of the most detailed findings come from smaller subsets of just 15 to 60 devices. The team is honest about their limitations throughout, which is part of what makes the findings worth taking seriously.

The Facade That Fooled Everyone, Including the Person Living It

Here is one of the most important findings that survivors most need to hear. Researchers applied a measure of suicide risk language usage, clinically validated at Harvard, to two layers of each phone:

• Private notes and searches (Private layer)
• Texts and calls a person actually sent to others (External layer)

The private layer showed a real, measurable climb in risk language in the year before death.

The external layer barely moved.

People experiencing the worst of it were, on average, keeping their outward communication close to normal while their private digital behavior told an entirely different story. Anxious language and agitation showed the strongest rise, ahead of sadness, which is not what most of us would have guessed.

This is not a new theory. It is the first large-scale digital confirmation of something clinicians have described for years, Dr. Igor Galynker’s suicide crisis syndrome, a state marked by frantic hopelessness under a surface that can look composed, and psychologist Richard Heckler’s suicidal trance, a narrowed state where a person can present as recovering even as the crisis tightens.

Many suicide loss survivors describe watching someone seem to calm down right before the end. The Black Box Project did not discover this pattern. It measured it, on real phones, at a scale no interview study could reach.

We lost our son John to suicide in 2009. Phones existed then, but they did not yet hold the diaries, the confessionals, and the late-night searches that a phone holds now. I have wondered more than once what a device from that year could have told us, and honestly, I am not sure I would have wanted to know in real timeKnowing does not always feel like power. Sometimes it just feels like a second, harder loss.

If you are now tempted to unlock your person’s phone because of what this research describes, that instinct makes complete sense, and there is a post on a full guide on accessing a device after suicide loss for that decision. I want to be honest with you here, the way I am with people in our group.

That does not mean you should not look. It means the decision deserves the same care you would give any decision you cannot take back. And if you have already looked, and you are sitting with something you cannot put down, that is not something to carry alone. A therapist or a support group meeting is a better place for that weight than your own mind at 2 a.m.

Whatever crisis language lives in that private layer reflects a narrowed, frightened state, not a more honest version of the person you lost. It is a record of their worst hours, not the whole story of who they were.

While we are talking about devices, most families also face a separate, practical question, what to do with the accounts and passwords a person leaves behind. There is also a post that is a guide on managing a digital legacy after suicide loss for when you get there.

The Ten-Week Shift: What Sleep and Late-Night Hours Revealed

Another finding from the study was that late-night phone activity between midnight and four in the morning climbed from a baseline near 30 percent of nights to just over 53 percent in the final week of life. The clearest turning point sat around ten weeks before death, when overnight activity began a steady climb that did not let up.

An earlier, smaller phase of the research has linked late-night activity to a much higher hazard of suicide, though the researchers flag that figure as preliminary.

Not everyone showed the same overnight pattern. About a third of devices showed heavy late-night use throughout, a third showed a sharp late increase, and the rest showed very little, even near the end. That variation matters. Insomnia and 3 a.m. scrolling were not universal warning signs anyone could have caught in the moment. They were real in aggregate, not something one person watching one phone could have detected.

Still Fighting: What the Cycles of Searching and Reaching Out Show

The research team identified 106 specific, timestamped moments of elevated risk, spread across 42 percent of the devices that they had analyzed. They were actual notes, searches, and web history entries that human reviewers confirmed reflected real distress, sorted into categories like researching methods, reaching out for help, and feeling like a burden to others.

What the researchers found around those moments is important. Browsing behavior in the days around a known risk moment often cycled back and forth, a search for a harmful method, followed not long after by a visit to a mental health resource, a support forum, or a faith-based site, then sometimes back again. That is not confusion. It is ambivalence happening in real time, a mind still looking for a way out that does not end in death.

This matches something survivors have heard from us before if you have read about why suicide is not about a desire to die. Psychiatrist Edwin Shneidman’s research into suicide notes described the same pull, a person in unbearable pain who wanted the pain to end far more than they wanted to die. The cycling this study found on actual phones is that same pull, made visible in search bars instead of only in theory.

The team also found that once planning-related searches appeared, they often escalated within days. That sudden speed is hard to read about. The weeks-long warning window many families imagine often was not the window that actually existed. The crisis compressed faster than almost anyone watching from outside could have tracked, and that pattern matches decades of separate research on the timing of suicidal crises, where some attempt survivors have reported windows of minutes, not days, between deciding and acting. Fast does not mean the suffering underneath it was fast too.

Money, Isolation, and the Quiet Signals Nobody Was Watching For

Financial stress turned out to be one of the most stable signals the researchers found. Where anger and general negative language rose and fell without reaching significance, financial distress language climbed steadily across the final year and accelerated sharply in the last month. It did not spike once and vanish. It sat there, getting stronger for almost a year.

Communication patterns shifted too, though not in one direction. Some people reached out more in their final weeks, increasing calls and messages in ways that may have reflected help-seeking. Others pulled back, narrowing their circle to fewer core contacts. Both patterns clustered around the same six-week and two-to-three-month thresholds before death, then often reverted to normal in the final week, its own quietly devastating detail.

The paper is also honest about what it did not find. Email interests, shopping habits, and fitness-tracker activity showed no meaningful pattern, which protects survivors from a specific trap, the urge to reread your person’s shopping history hunting for a clue that was never going to be there.

Browsing habits otherwise stayed remarkably stable, though visits to financial planning sites quietly rose, the kind of forward-looking browsing that can look like responsible planning right up until it is something else entirely.

If this section is bringing up how exhausting it was to watch someone pull away and reach back out, over and over, take a breath. That whiplash was real.

This Was Never Only a Veteran’s Story

Stop Soldier Suicide exists because veterans and service members die by suicide at a substantially higher rate than the general population, and that mission shaped who first donated devices. But the organization has been direct that the device pool now includes people from every kind of background, not only military service. SAVE, a national suicide prevention organization that partners on the project’s survivor-facing side, confirms devices come from individuals with all lived experiences, not only those with a military connection.

That distinction matters no matter what kind of loss brought you here. In fifteen-plus years of facilitating meetings at SOS Madison, I have heard versions of “they seemed OK,” the late-night phone use, the sudden financial worry, the pulling away then reaching back out, from people who lost a spouse, a teenager, a parent, a sibling, a close friend. While the majority of the people in our group have not lost a military member,the pattern showed up anyway.

None of the mechanisms this research describes are military-specific. A narrowed, crisis-state mind does not check someone’s employment history first. Suicide crisis syndrome, the suicidal trance, and the ambivalence Shneidman documented decades ago were all described using civilian research long before this study analyzed a single phone.

What the Black Box Project adds is scale and digital proof, not a new population where the crisis works differently.

What This Means for the Guilt You Have Been Carrying

If you have spent months or years running the “should have, could have, would have” loop, replaying the last conversation, the last text, the last ordinary Tuesday, this research is not going to make the loop stop on its own. But it can change what you feed into it.

Here is something to think about:

• The digital record of internal crisis this study found lived almost entirely in private notes and things never sent to another person.
• Your person’s outward communication, the calls and texts you actually had access to, stayed close to steady for most people in this study.

I have written about hindsight bias, the mind’s habit of assuming the signs must have been obvious simply because we now know how the story tragically ends. This research gives that idea actual measurement instead of only intuition. The signs were often real. They were also, for the person standing next to your person that week, functionally invisible.

There is a difference between guilt and responsibility, and it deserves to be said clearly rather than only felt. Guilt says I should have caught this. Responsibility asks whether you actually had access to the information that would have let you catch it.

I have also written more on guilt versus responsibility after a suicide loss if that distinction is one you are still working through. For most families in this study, the honest answer is that the state your person’s brain was in narrowed what they could put into words and who they could say it to. That narrowing is not the same as deception. It is what this kind of crisis does to a person’s ability to reach out, whether or not they wanted to.

There is rarely one single cause behind a suicide. For years I have used the image of a glass slowly filling with water to describe how risk factors accumulate until one final drop overflows it, and this study is really a close-up photograph of that glass filling, mostly out of view.

The last drop was never the whole story.

Where Suicide Prevention Goes From Here

Most suicide screening depends on what a person is willing to say out loud, in a doctor’s office, to a friend, on an intake form. This research suggests that approach will keep missing exactly the people most at risk, since this kind of crisis narrows what a person can disclose, not because anyone chooses to hide it. That is hard to hear if you have ever asked someone directly and been told yes. It is also not a failure on your part. If you have ever found yourself an unexpected point of contact for someone else’s crisis, there is a post that is a guide for what to do when someone confides in you, built around the ask, be there, keep safe, help connect, and follow up steps the National Institute of Mental Health recommends.

Sleep, financial stress, and social withdrawal matter here because they are not private the way a person’s inner thoughts are. A sleep disruption can show up to a partner. A money problem can show up to a bank. Pulling away from friends can show up to almost anyone paying attention. Prevention tools built around these more visible signals could reach people sooner than waiting for someone to say the hardest sentence out loud.

If reading this has you thinking about someone else in your life right now, that instinct makes sense. Losing someone to suicide leaves a lot of survivors watching everyone else more closely than they used to. That watchfulness is not shameful, and it is not sustainable by checking someone’s phone. If you are genuinely worried about someone, the sturdier path is connecting them to support, a real conversation, a professional, a crisis line, rather than monitoring alone.

The American Foundation for Suicide Prevention and the Suicide Prevention Resource Center both frame suicide loss survivors as uniquely positioned to reshape prevention, work they call postvention, using what survivors and researchers learn after a loss to prevent the next one. I have written more on why postvention functions as prevention.

Some families take that further by donating a device to the Black Box Project, a decision that is entirely personal. It is also a decision some sit with for a long time and decide against, and that is a complete answer too. Handing over a phone means letting strangers, however careful and respectful, read the most private hours of a life you are still grieving. Feeling uneasy about that is not a failure of generosity. It is a reasonable response to a genuinely hard ask.

Questions Survivors Ask About the Black Box Project

• Does the Black Box Project mean I should have seen the signs? No. The strongest signs of crisis lived in private notes and searches, not the calls and texts most family members had access to. A steady-looking outside matches what this research found in most cases.
• Is the Black Box Project only about veterans and service members? It started there, since Stop Soldier Suicide’s mission is veteran suicide prevention. The organization has since confirmed the project includes devices from people with all kinds of lived experience, and the patterns found are not military-specific.
• Can a family donate a device to the Black Box Project? Yes, and it is entirely voluntary. Families loan a device to the research team, who handle it with forensic procedures and return it within about ninety days. There is no timeline, and no family is ever pressured. Families who want to start that conversation can reach out directly at bbxproject.org/families.

Reading a study like this can stir up a lot, the guilt, the wondering, the strange grief of learning your person may have been fighting harder and more privately than you ever knew. If that is where you are right now, you are not doing this wrong. Feeling shaken by this research is not a sign you have more work to do. It is a sign you cared enough to keep learning about someone you lost.

What I hope you carry out of this is something closer to relief than to feeling like there is more evidence against yourself. For many people in this study, the signs that mattered most were never visible in the places family and friends actually had access to look. If that matches your experience, the searching you have done since, the replaying, the wondering what you missed, was probably never going to turn up an answer that was sitting in plain sight.

I also know that is not everyone’s story. Some of you watched something real happen and did not know what to do with it in time. Some of you asked directly and were told not to worry, by the person, by a doctor, by your own hope that you were overreacting. Some of you did everything anyone could reasonably have done, and it still was not enough.

Before his suicide, we knew that our son John had been struggling for months, ever since his bipolar diagnosis. We had arranged for doctors, therapists, inpatient and outpatient therapy. Early in his crisis, right after that diagnosis, we believed that was when he was at the greatest risk. It was not. After that initial period, the external layer was positive and reassuring, telling us he was on a path toward healing. Back in 2009, devices were fairly limited in what they could do. His phone could make calls, send text messages, and play Snake, and I did have access to his desktop computer’s internet searches. After he died by suicide, a review of his computer showed that in the final hours of his crisis, those searches turned dark and dangerous, in a way none of us saw coming. We now know that the reassuring external layer was not the truth, and we lost him to suicide. Our glimpses into his private layer after his suicide showed us that the inner turmoil was real, and not what he was showing to everyone around him.

Seeing a sign and having the tools to read it correctly in the moment, with no hindsight, are two very different things.

That gap is not a character flaw. It is the ordinary limit of being human and afraid for someone you cared for deeply, and this research does not erase that any more than it erases the loss itself.

Whichever version of this is yours, the weight you have been carrying was never proof of what you failed to do.

Posts You May Also Like

• Suicide Crisis Syndrome: Why They Couldn’t Just Tell You – Why the outside can look calm.
• Suicidal Trance: What It Means for the Questions You Still Carry – A narrowed state that mimics calm.
• Hindsight Bias: Healing the Pain of “Could Have Known” After Suicide Loss – For the belief the signs were obvious.
• Guilt vs Responsibility After Suicide Loss: You Are Not to Blame – Feeling guilty is not being responsible.
• Is Suicide Preventable? Not Always, and Never Your Fault – Takes the guilt this research surfaces and works through it directly, including the attempt-survival numbers and the difference between guilt and responsibility.
• Suicide Loss Survivors: 6.6 Million New Grievers Every Year – Why survivor-driven research is itself prevention.
• Accessing a Device After Suicide Loss: Before You Unlock It – Whether, and how, to look.
• Digital Legacy After Suicide Loss: What to Do With Accounts – Managing accounts once you are ready.

PRINTABLE GUIDE PDF

A two-page PDF guide has been generated for survivors to print, save, or share.