Protecting Truth in a Synthetic Age.

The Global Authentic Memory Initiative builds open tools and standards to prove authenticity and integrity of digital witness material — designed for archives, memorial sites, media, researchers and civil society.

Get in Touch How Verification Works

Why a Global Initiative for Authentic Memory?

Synthetic media makes it easy to imitate voices and faces — and easier still to wave away authentic evidence as fabricated. For remembrance, education and journalism, this erodes trust where it matters most.

GAMI focuses on a neutral, open, long‑term proof layer that makes provenance and integrity verifiable without centralising content.

What We Protect

Video and audio interviews, written testimonies, digital collections from archives and memorial sites — content whose credibility matters to future generations.

What We Are Not

We do not certify what is true. We provide tools to show who signed what, when, and how that claim can be checked.

The Threat: The Liar’s Dividend

As people learn that images, audio and video can be fabricated, it becomes easier for bad‑faith actors to dismiss authentic evidence with a simple assertion of fabrication.

Further reading: Chesney & Citron (2019); Schiff (2025).

A people that no longer can believe anything cannot make up its own mind… And with such people you can then do what you please.

— Hannah Arendt

Why Now?

Last witnesses

As firsthand witnesses pass away, it becomes much harder to counter synthetic narratives.

Closing window

Unaided human judgment will soon struggle to tell real from synthetic media.

Anchor today

Anchoring authenticity now preserves trust for the decades ahead — across platforms and institutions.

Short Explainers

Clarifying the scope of authenticity infrastructure and why independence matters.

Deepfakes and Authenticity

Generative AI tools are constantly improving, producing better synthetic images, audio and video every year.

Worryingly, AI‑generated images purporting to depict Auschwitz already circulate online. Such fabrications blur the line between documentation and invention, retraumatize survivors and their families, mislead learners encountering history for the first time, and erode trust in authentic archival records.

Authenticity infrastructure does not stop deepfakes from being created. What it can do is protect authentic material from being wrongly dismissed as inauthentic, highlight when content has no recorded, verifiable history, and support journalism, education and research in mixed synthetic/real environments.

Authenticity vs. Truth

Authenticity concerns verifiable provenance and integrity — not adjudicating meaning or morality.

An authenticity layer can show who signed a file, when and where it was anchored, and whether its bits have changed since. It cannot decide whether every statement in a testimony is factually correct, morally acceptable, or historically representative.

Why Independence From Any Single Institution Matters

Trust in individual institutions changes over time. Archives may be restructured, hacked, under political pressure or simply disappear. Even reputable organizations can be accused of bias.

By anchoring proofs across multiple, tamper‑evident systems — and by designing the registry to be mirrorable by many parties — authenticity checks remain possible even if one institution is compromised and can be trusted, in principle, by people who distrust any particular government or vendor.

This distributes risk, reduces single points of failure, and helps keep verification available across generations and jurisdictions.

Who It’s For

GAMI’s authenticity layer is designed for multiple domains that rely on credible records.

Archives & Memorial Sites

Preserve trust in collections at internet scale without centralising content.

Media & Platforms

Indicate verifiable provenance and integrity across distribution and reuse.

Education & Research

Distinguish authentic sources in mixed real/synthetic environments.

Civil Society & Law

Reduce plausible deniability and support accountability processes.

Scenarios: Where Authenticity Proof Matters

Plausible, policy‑relevant cases illustrating risks and how neutral proofs of existence and integrity can help.

Future Denial Campaign

Synthetic look‑alikes cast doubt on established testimonies.

▾

In the coming years, coordinated networks claim that well‑known Holocaust testimonies are AI‑generated. Short synthetic clips mimic the visual style of archival interviews and introduce subtle contradictions. The tactic builds on patterns of denial and distortion already documented on major platforms.

How authenticity proof helps: When testimonies are cryptographically anchored upon ingestion, independent tools can show that a given recording existed in the same form long before the campaign and has not been altered.

The Hacked Archive

Quiet tampering and metadata drift over time.

▾

A large audiovisual collection suffers a sophisticated intrusion. A small number of files are altered or removed; some metadata is changed to make sensitive items appear newer than they are. Years later, investigators see inconsistencies but lack a shared external record for comparison.

How authenticity proof helps: External, tamper‑evident anchors let investigators compare current files to historic proofs and establish which versions existed, when, and in what form.

The Researcher in 2080

Differentiating originals, adaptations and manipulations at scale.

▾

By 2080, historians study tens of thousands of testimonies across formats and languages. Reconstructing the institutional and technical history of each collection — platform migrations, CMS changes, publication policies — is often impossible decades later, yet essential for serious scholarship.

How authenticity proof helps: A shared authenticity layer answers a narrow but critical question: did this exact file exist at time X, and has it remained unchanged since?

The 17‑Year‑Old on Social Media

Authentic clips and synthetic look‑alikes side by side.

▾

A student encounters survivor testimony and synthetic witnesses in similar styles. Without reliable provenance, both appear equally credible. Surveys show knowledge gaps and that social media is often the first contact with denial and distortion.

How authenticity proof helps: Portable, platform‑independent proofs let platforms and educators indicate that a clip comes from a recognised institution and verifiably existed in its current form long before current debates.

The Plausible Deniability Defense

Plausible deniability stalls accountability.

▾

In inquiries or courts, some parties no longer try to prove that evidence is fabricated; they argue that, given today’s AI, it could be. This tactic — the liar’s dividend — is already noted in scholarly and policy debates.

How authenticity proof helps: Evidence anchored across independent systems at the time of recording is harder to dismiss wholesale. It supports reasoned scepticism without enabling blanket doubt.

How Cryptographic Protection Works

The technical architecture is under active development. Based on our current design work, it is likely to take a form similar to the following:

Create a Cryptographic Fingerprint

Institutions keep their files; a secure hash uniquely represents a file’s exact state.

Anchor and Record

Bind the hash to trustworthy time and write the proof to an open, append‑only registry mirrored across independent, tamper‑evident systems.

Verify Years Later

Anyone can re‑compute a file’s hash and compare it to the registry to confirm integrity and earliest known existence.

The goal is a neutral, long‑term mechanism: a minimal, durable proof of existence and integrity — without ever centralising the files themselves.

Governance & Ethics

Neutral governance: independent advisory board across archives, cryptography and memory institutions.
Open by default: public spec and reference code; mirrorable registry format.
Privacy & Personality Rights: compatible with sensitive material; only non‑reversible cryptographic fingerprints are stored, never the underlying files.
Resilience: multi‑anchor design, offline verification paths, and clear threat modeling.

From Idea to Infrastructure

Phase 1 — Foundations

Pilot Projects With Partners

Co-design with archives, memorial sites, journalists, legal and cryptography experts.

Phase 2 — Opening

Public Registry & Tools

Freely usable tools and interfaces.

Phase 3 — Stewardship

Long‑Term Structure & Governance

Durable governance and international cooperation.

Frequently Asked Questions

Short answers to common concerns about scope, privacy and technical design.

Can a hash reveal the original file?

▾

No. A cryptographic hash is a one‑way fingerprint. It allows integrity checks without reconstructing the original content.

What if a ledger disappears or is compromised?

▾

Proofs are anchored across multiple, independent, tamper‑evident systems and the registry format is mirrorable. Verification can work even if one system fails.

How is this different from watermarking?

▾

We don’t alter the content. We anchor external proofs of integrity and time, enabling verification even when files leave controlled environments.

Do you upload or store our files?

▾

No. Institutions keep their files. Only non-reversible fingerprints are stored on decentralized systems.

How does this relate to C2PA?

▾

We’re complementary. GAMI focuses on long‑term, institution‑independent proofs and integrity checks; C2PA addresses provenance within platformed workflows.

Could this enable tracking or surveillance?

▾

We minimise metadata, support pseudonymous anchors and keep the spec public for audit. Ethical safeguards are part of governance, not an afterthought.

We’re Exploring — Your Input Matters.

GAMI is in active exploration and research. If you’re interested in contributing — conceptually, institutionally or financially — we’d love to hear from you.