About the POW Investigative Project

Have information? Email contact@powinvestigativeproject.org, use the contact form or Signal PIP.21. Information can be shared in confidence. No payment is offered.

In this section

  • Team — the expert volunteers, editorial team and pro bono legal advisers behind the project.
  • News — POW/MIA News, on official accounting, litigation, foreign archives and the family organizations, and News About PIP, on stories that name the project or our investigators and advisers.
  • Frequently Asked Questions

About the project

The POW Investigative Project seeks answers on the fates of Americans reported or believed captured by the communists during the Korean, Cold and Vietnam Wars, but never returned or accounted for. We are an all-volunteer group of ex-government officials, investigative journalists, subject-matter experts and family members.

PIP, Inc. is a 501(c)(3) private operating foundation — an investigative reporting and multilingual crowdsourcing platform built to help resolve one of the most enduring unanswered questions in American history.

More than 9,000 U.S. servicemembers remain unaccounted for from those three conflicts, according to the Defense POW/MIA Accounting Agency (about 7,353 from Korea, 1,565 from Vietnam and 126 from the Cold War, as of Sept. 18, 2026). Many of their families are still seeking answers.

Our purpose

PIP exists to do three things:

  1. Preserve evidence and build the historical record. Declassified intelligence files, contemporaneous news reporting, family papers and firsthand testimony are scattered, fragile and in some cases still classified. We gather them, cite them and publish them.
  2. Get answers. We believe solutions to the fates of many missing Americans exist in the records of foreign governments, and in the memories and possessions of former officials and ordinary citizens in countries where U.S. prisoners were reported held. We publish in multiple languages to reach them.

How we work

We identify promising leads and clues — including material we have obtained by forcing the declassification of U.S. intelligence records — and publish profiles of missing Americans for whom information is sought. We use multilingual crowdsourcing to reach potential witnesses abroad, and we retain investigative reporters and freelance researchers to pursue specific questions.

PIP is all-volunteer. We are a group of volunteer ex-government officials, investigative journalists, veterans, subject-matter experts and POW/MIA family members. Brant H. Dveirin and his firm, Lewis Brisbois, provide the project’s legal support pro bono.

How AI expanded this work

The investigation is nearly forty years old. What is new, since 2026, is that a machine can hold all of it at once.

The kind of connection that resolves a case

An intelligence report in one file. A loss location in another. A physical description in a third — a height, a scar, a mole on the throat, a missing finger. A camp report describing an American who is never named. A transliterated surname in a foreign document that no English-language search will ever return.

Not one of those is worth anything on its own. Put together, they are an identification.

That has always been the work, and it has always been done by hand: one researcher, one folder at a time, hoping to recall on a Tuesday something read three years earlier in a different archive. Almost every case on this site sat unresolved not because the evidence was missing but because the pieces were never in the same room.

AI-assisted research compares everything against everything. Every name, date, rank, unit, aircraft type, place name, case number, camp designation and physical description, in every document, against every other — including across languages and transliterations. It does in an afternoon what a researcher could hunt for months without ever reaching, and it finds the connections nobody thought to look for, which are usually the ones that matter.

What it reads

Far more than one archive:

  • The Library of Congress POW/MIA database — the declassified record assembled around the Senate Select Committee on POW/MIA Affairs and since expanded: about 164,000 records (163,649 at the Library’s last published count), which by our count runs to close to a million pages. Intelligence reports, debriefs, casualty files, camp rosters, referral documents, hearing exhibits. Most of it has been public for thirty years and read by almost no one.
  • The project’s own declassification work — tens of thousands of pages that PIP has forced into the open through Freedom of Information Act requests, Mandatory Declassification Review and litigation. Much of it has not been published anywhere we know of.
  • Foreign archives — Russian, Chinese, Vietnamese, Polish, Czech, Hungarian, Lithuanian, Ukrainian and Korean holdings, read in the languages they were written in rather than waiting in a translator’s queue.
  • The non-English internet — the part of the record no English-language search reaches at all. Russian veterans’ forums and unit histories, Chinese and Korean local press, Vietnamese provincial and memorial sites, regional genealogy databases, obituaries, and memoirs published only in the original language. Several of the leads on this site began there.
  • Books, memoirs and the project’s own case files — decades of accumulated correspondence, interview notes, family material and prior research, now searchable as one body rather than forty years of folders.

Documents that were never machine-readable

Government responses to declassification requests routinely arrive as photographed pages with no text layer — ninety pages of images that cannot be searched at all, and that in practice nobody reads twice. Those are now transcribed and searchable. One such file, a response on East Bloc and Hungarian reporting about American prisoners, was read in full this year for the first time since it was released.

Maps and visual analysis

Once the record is machine-readable, it can be plotted. Loss locations, reported camp sites, transfer routes, sightings, interrogation locations and burial claims can be placed on the same map and the same timeline — so that a 1952 sighting sitting forty miles from a 1951 loss becomes something you can see rather than something you have to hold in your head. Patterns buried in fifty years of prose become a shape on a page.

Photographs and imagery

A great deal of this record is not text. It is pictures — compounds photographed from the air, men in camps, identification cards in the display cases of foreign museums. AI reads a picture much as it reads a document: what is physically in the frame, the writing in a corner of it, the same face in two photographs taken years apart, checked against every other document in the collection at once.

The practical change is simpler than it sounds. When a report gives a place and a date, we can now go and find out what overhead imagery exists for that place at that time. Declassified American satellite photography from 1960 to 1984 is public, searchable and free, and a great deal of it has never been looked at by anyone outside the government.

Take Maj. Henry M. Serex, lost over Quang Tri Province on Easter Sunday 1972 aboard the EB-66 known as Bat 21. His escape-and-evasion authenticator — the number that would prove a signal came from him and not from someone using his name — was 72TA88. On June 5, 1992, U.S. overhead imagery of a field beside a prison at Dong Mang, in Quang Ninh Province, was read as carrying three sets of symbols: SEREX, 72TA88 and GX2527. Seventeen years earlier, on July 1, 1975, American analysts produced a symbol readout of the same place, recording rooftop markings.

So there is a place, and there are two dates. If those particular frames have been released, we do not have them. But the 1975 date falls inside the declassified window, which means imagery of that place from those years can be searched, obtained and read now, by us or by anyone else. The 1992 frame is a records question rather than an imagery one. Either way there is a specific place and a specific date to work from, which is what cases like this have usually lacked.

What this cannot do matters too. It cannot add detail that is not in the picture, it cannot date a photograph on its own, and a reading is a description to be checked rather than a finding. But the question can now be put, and put precisely.

If you hold photographs bearing on any of these cases, or you have worked in imagery analysis and photo interpretation, we would like to hear from you.

Appeals in a new language, the same day

When a lead points to a town in Ukraine, a camp in Kazakhstan or a province in Vietnam, we can now write and publish a request for information in that language within minutes. That used to mean weeks of turnaround and thousands of dollars in professional translation — which for a small charity meant, in practice, that most appeals were never made at all. Appeals now stand on this site in more than a dozen languages.

That is not a convenience. The people who can still answer these questions are abroad, and they are in their eighties and nineties. Every week not spent waiting on a translation is a week spent looking.

What has not changed

Nothing is published on a machine’s say-so. Every finding on this site is checked against the primary document before it appears. A correlation is described as a correlation and never as an identification. Where the record contradicts itself, we record the conflict rather than resolve it quietly. Where a claim rests on a single source, the page says so. Our senior advisors, including former government experts on these cases, judge what is credible — as they always have.

The tool is faster. The standard of proof is not lower.

What it has already produced

And the response that matters more than any of it. The granddaughter of a missing U.S. Navy aviator, shown foreign-language material this research turned up on her grandfather’s case:

“This is a wonderful tool that hopefully can get us closer to the truth… I have never seen any of that from DPAA.”

That last clause is the point. Not that the machine is clever — that it reaches records the families were never shown.

Does the Pentagon not resolve these cases?

After years of controversy involving the Pentagon’s POW/MIA office — now the Defense POW/MIA Accounting Agency (DPAA) — Congress required the organization to sharply increase the number of POW/MIA remains it identifies each year. The office now largely focuses on identifying remains that have been in U.S. control for decades — unidentified men from Korea buried in Hawaii, sailors entombed at Pearl Harbor, and remains long ago provided by communist governments. PIP believes that identifying these men is a crucial, high-priority task, as is recovering the remains of as many other missing Americans as is feasible in communist countries. We commend the military men and women who search for the remains of U.S. POW/MIAs in challenging, uncomfortable and sometimes dangerous land and maritime locations.

That is not the focus of PIP. We exist to investigate reports of Americans secretly imprisoned or held back by communist forces. In many cases DPAA now contradicts the U.S. government’s own declassified records to deny reports of these men. During the Korean War, for example, a unit composed of American military intelligence and CIA officers “confirmed” the existence of multiple secret POW camps in China holding U.S. prisoners. No Americans returned from those camps, but today the Pentagon does not include those locations, or those men, in its analytical and recovery efforts. Many outside observers — including former government officials and a number of POW/MIA family members — believe the U.S. government has effectively set most such cases aside.

Our government also continues to keep critical POW/MIA information classified, some of it dating back to 1951, despite years of Freedom of Information lawsuits and requests that we and others have pursued. See our FOIA lawsuits.

U.S. officials and former officials have said Moscow, Beijing, Hanoi, Vientiane and Pyongyang hold records they have not provided that could resolve the fates of many missing Americans.