The Ledger of Truth: A Signed Web for Zero‑Trust Media

1) The core problem: the detection arms race is futile

Most proposals to fight deepfakes lean on AI/System detection (blink rates, audio spectrogram's, pixel artifacts). But half life of such detection models is quite short. Every AI model breakthrough shrinks their half‑life. We can’t rely on content analysis to tell us if something is real. We must validate the source and the chain of custody. In other words: content provenance.

2) Proposal: a public, “signed” web

SSL for web functions on the premise of zero trust. Meaning, every website is untrustable until proven to be real. And a SSL certificate is a digital signature from a trusted authority who guarantees the authenticity of the website.

But executing the same for public assets is a challenge purely because of the vast amount of content we have today.

A public record (like a blockchain) where content creators publish verifiable signatures on their content. This changes the default from "assume real until proven fake" to "assume unverified until proven authentic."

3) Technical architecture

Layer 1: The Identity Bridge (real‑world → digital)

• Create your identity: Generate a unique digital signature (like a username and password, but more secure) or key pair (public, private keys).

• Get verified: Confirm your identity through an official government service (like eIDAS in Europe or DigiLocker in India). This links your digital signature to prove that you're a real person from a specific country—without revealing your name publicly. The system uses privacy technology (zero‑knowledge proofs) so you can prove things like "I'm verified," "I'm over 18," or "I'm from Country Y" without exposing who you actually are.

Layer 2: The Signing Process (hashing)

• Hash media with a cryptographic function (e.g., SHA‑256) to get a unique fingerprint.
• Sign the hash with the verified private key.
• Record the content hash + signature + key attestation on the public ledger.

Layer 3: The Consumption Layer (browser/app)

Platforms and browsers integrate a simple check: when media loads, compute the hash locally and query the ledger:

• If there’s a verified signature match, show a green check with the signer name or entity and verified country.
• If not, show a neutral “Uncertified source” badge. The absence of a signature doesn’t declare something “fake”, it declares “origin unknown.”

4) Scenarios

A) Deepfake politician video

Today: a deepfake video spreads, people panic, and fact-checkers spend hours investigating. With a signed web: the president's official account hasn't signed this video. Your browser shows "Uncertified." People dismiss it quickly.

B) News journalism

A reporter signs footage on upload. If someone edits and re‑uploads, the hash changes, breaking the signature. Viewers instantly see it’s not the original.

5) Critical challenges and mitigations

• Privacy trade‑offs: Linking to government‑grade identity raises surveillance concerns. Mitigation: ZKPs + selective disclosure. The ledger can attest to uniqueness/citizenship without exposing names unless the signer opts in (e.g., public officials, news orgs).
• Stolen keys: If a private key is compromised, an attacker can sign fakes. Mitigation: Hardware security (HSM/secure enclaves), multi‑device approvals, fast revocation/rotation, and a visible “last rotated at” signal.
• The oracle problem: This proves origin, not truth. Verified people can still lie. The value is accountability and provenance. Bad actors can be stripped of trusted status and held responsible.

6) Why this is practical now

• Mature building blocks: modern signature schemes, commodity secure enclaves, and standards work on content provenance.
• UX can be simple: badges by default, details on click.
• Interop first: Start with public figures, news orgs, and high‑risk media; let long‑tail creators opt in over a period of time.

7) Closing

We’re past the “seeing is believing” era. The default posture for media should be zero‑trust: “verifying is believing.” A signed‑web standard—anchored in sovereign identity attestation and an immutable public record—creates a whitelisted path where human accountability is the prerequisite for trust.

In a follow-up post, let’s dive deeper and implement a proof of concept together.

References and further reading