A guest contribution from the Global Authentic Memory Initiative (GAMI), Munich, Germany
When memorial sites and educational institutions published a joint open letter against AI-generated Holocaust distortions in January 2026, they pointed to a danger that goes beyond the fabricated images themselves: confronted daily with convincing fakes of supposed scenes from the Holocaust, users have begun to doubt even authentic historical documents. Whether emerging as a by-product of a changing media landscape or deliberately perpetrated, the erosion of public trust in historical evidence is becoming antisemites’ cheapest weapon.
The antidote to doubt is verifiable evidence. For newly created content, provenance standards such as C2PA now attach verifiable origin information the moment a file is produced. The digitised holdings of archives, memorials and museums, however, predate these standards. As certified content becomes the norm online, real historical material, including survivor testimony, risks falling From thinto a grey zone, unable to provide proof of authenticity in a digital environment that now expects it.
The Global Authentic Memory Initiative (GAMI), a non-profit organisation based in Munich, is building open infrastructure to close this gap. The method is deliberately simple: an institution computes cryptographic fingerprints (SHA-256 hashes) of its digital holdings on its own servers. No file ever leaves the institution, and nothing is modified. The institution signs these fingerprints, which are then anchored in open, tamper-evident public infrastructures. The result is a durable proof that a given file:
- existed at a documented point in time (Existence),
- carries the verifiable signature of the holding institution (Provenance),
- and has remained unchanged since (Integrity).
Anyone can check these properties, free of charge and without an account, at verify.authenticmemory.org. The tooling is open source, and the proofs remain valid independently of GAMI itself.
A cryptographic proof cannot show that a photograph taken in 1945 depicts what it claims to depict. What it can show is that this exact file existed, and was vouched for by a named institution, no later than 2026. Thus, even decades later, one can rule out that the photograph was created by more advanced AI tools developed after 2026.
If future generations are to differentiate real images from fake ones, the time to establish the proof is now.
GAMI’s idea is to make use of the special window in the development of AI-generated media that is the current moment: A time when the threat it poses is obvious, yet technically indistinguishable deepfakes do not yet exist. If future generations are to differentiate real images from fake ones, the time to establish the proof is now.
This is why GAMI launched a pilot phase to test the approach under real archival conditions. In June 2026, the first stage was successfully completed with the Dachau and Flossenbürg Concentration Camp Memorials: the first batches of archival records – including survivor interviews and accompanying metadata – were signed and anchored on the Bitcoin blockchain within days, and each record can now be checked through the public verification page. The development of the technical foundation is funded by Stiftung Erinnerung, Verantwortung, Zukunft (EVZ). Workflows, cost data and technical specifications will be published openly, and the long-term aim is a shared standard governed by the participating institutions, interoperable with C2PA. An extended pilot cohort is now being assembled, and GAMI is actively looking for institutions to take part.
Get involved
The strongest proofs are established now. GAMI aims to do exactly that for at-risk records within the next few years. This is why the initiative invites archives, memorials and museums to join the ongoing pilot phase and the standards discussion around it.
Interested institutions are encouraged to get in contact via info@authenticmemory.org. Documentation and the public verifier are available at authenticmemory.org.
A guest contribution from the Global Authentic Memory Initiative (GAMI), Munich, Germany
When memorial sites and educational institutions published a joint open letter against AI-generated Holocaust distortions in January 2026, they pointed to a danger that goes beyond the fabricated images themselves: confronted daily with convincing fakes of supposed scenes from the Holocaust, users have begun to doubt even authentic historical documents. Whether emerging as a by-product of a changing media landscape or deliberately perpetrated, the erosion of public trust in historical evidence is becoming antisemites’ cheapest weapon.
The antidote to doubt is verifiable evidence. For newly created content, provenance standards such as C2PA now attach verifiable origin information the moment a file is produced. The digitised holdings of archives, memorials and museums, however, predate these standards. As certified content becomes the norm online, real historical material, including survivor testimony, risks falling From thinto a grey zone, unable to provide proof of authenticity in a digital environment that now expects it.
The Global Authentic Memory Initiative (GAMI), a non-profit organisation based in Munich, is building open infrastructure to close this gap. The method is deliberately simple: an institution computes cryptographic fingerprints (SHA-256 hashes) of its digital holdings on its own servers. No file ever leaves the institution, and nothing is modified. The institution signs these fingerprints, which are then anchored in open, tamper-evident public infrastructures. The result is a durable proof that a given file:
- existed at a documented point in time (Existence),
- carries the verifiable signature of the holding institution (Provenance),
- and has remained unchanged since (Integrity).
Anyone can check these properties, free of charge and without an account, at verify.authenticmemory.org. The tooling is open source, and the proofs remain valid independently of GAMI itself.
A cryptographic proof cannot show that a photograph taken in 1945 depicts what it claims to depict. What it can show is that this exact file existed, and was vouched for by a named institution, no later than 2026. Thus, even decades later, one can rule out that the photograph was created by more advanced AI tools developed after 2026.
If future generations are to differentiate real images from fake ones, the time to establish the proof is now.
GAMI’s idea is to make use of the special window in the development of AI-generated media that is the current moment: A time when the threat it poses is obvious, yet technically indistinguishable deepfakes do not yet exist. If future generations are to differentiate real images from fake ones, the time to establish the proof is now.
This is why GAMI launched a pilot phase to test the approach under real archival conditions. In June 2026, the first stage was successfully completed with the Dachau and Flossenbürg Concentration Camp Memorials: the first batches of archival records – including survivor interviews and accompanying metadata – were signed and anchored on the Bitcoin blockchain within days, and each record can now be checked through the public verification page. The development of the technical foundation is funded by Stiftung Erinnerung, Verantwortung, Zukunft (EVZ). Workflows, cost data and technical specifications will be published openly, and the long-term aim is a shared standard governed by the participating institutions, interoperable with C2PA. An extended pilot cohort is now being assembled, and GAMI is actively looking for institutions to take part.
Get involved
The strongest proofs are established now. GAMI aims to do exactly that for at-risk records within the next few years. This is why the initiative invites archives, memorials and museums to join the ongoing pilot phase and the standards discussion around it.
Interested institutions are encouraged to get in contact via info@authenticmemory.org. Documentation and the public verifier are available at authenticmemory.org.