The Snowden archive is best understood not as a lost object but as a collection whose private custody outlasted its public release. A Libroot account published on 20 September 2026 says the final batch appeared on 29 May 2019 and that no news outlet, journalist, or institution has published another archive document since. That is a strong publisher assertion about the public record, but the supplied evidence contains no independent global inventory proving that no unpublished release occurred anywhere.
The archive’s distribution began before Edward Snowden’s identity was public. Libroot reports that Snowden sent Laura Poitras an encrypted file in March 2013, mailed a physical package in May, and provided keys to an archive called Pandora—described as containing about 50,000 documents—later that month. Poitras then distributed copies to several people and institutions to avoid what Snowden called a “single point of failure.”
Glenn Greenwald’s 2023 account independently confirms the broad scale and early distribution, although it comes from a participant rather than an external auditor. Greenwald recalls Snowden giving Guardian journalist Ewen MacAskill a memory stick containing tens of thousands of NSA and GCHQ documents, with still more material going to Poitras and Greenwald. He says the material subsequently supported reporting by The Guardian, The New York Times, and ProPublica. This corroborates the existence of a large, deliberately dispersed collection, but it does not verify the exact size or present condition of every copy.
The image accompanies Harper’s 2017 account of the archive’s distribution and custody, a subject discussed in the article’s account of its early dispersal.
The evidence also separates possession from access. Greenwald said in March 2019 that he and Poitras each retained full copies and that other individuals and institutions did as well. Poitras said in 2022 that the archive still existed and contained material of continuing historical importance. Those statements support the claim that copies survived beyond the last public batch. They do not show that all holders still possess working keys, that their copies are identical, or that either complete archive remains technically readable today.
Publication was always selective. Libroot reports that The Guardian held roughly 58,000 documents but published only about 30 before ending document releases in February 2014—approximately 0.05% of its stated holdings. The paper continued producing stories from the material afterward, including investigations into surveillance of journalists and GCHQ’s role in drone operations. The relevant shutdown was therefore not a complete end to archive-based reporting; it was an end to routine publication of underlying documents by that outlet.
Other partners followed different timelines. According to Libroot, Der Spiegel stopped publishing documents in January 2015, while The New York Times and ProPublica stopped in August 2015. The Intercept continued releasing material until March 2019, with the final batch appearing that May. The uneven schedule points to separate editorial decisions rather than one technical event that destroyed or disabled the entire archive.
Why those decisions accumulated remains unresolved. Libroot examines falling editorial attention, budgets, competing priorities, legal and national-security constraints, and the argument that older documents had diminished news value. It also notes that The Guardian faced substantial government pressure, consulted authorities on nearly all of its early Snowden stories, and destroyed the London computers holding its copy in July 2013—an act a former deputy editor later described as symbolic because copies had already moved to the United States. These facts establish real institutional pressures, but they do not prove which pressure caused any particular outlet to stop.
The archive’s historical significance also cannot be inferred from the small publication percentage alone. A limited set of documents can reveal major programs, while a larger unpublished collection may contain duplicates, marginal material, or additional evidence whose importance has not been assessed. The supplied sources establish neither that the released 1% was representative nor that the remaining 99% contained comparably important disclosures.
For readers, the central distinction is between disappearance and opacity. The evidence supports a chain of custody involving multiple journalists and institutions, continued private possession years later, and a final public release in 2019. It does not establish a current inventory, confirm that all copies are intact, identify who could authorize access, or explain comprehensively why publication ceased. Until such records appear, claims that the archive is gone, complete, suppressed, or still full of explosive revelations all outrun the evidence.