I'm looking at this as:
Content Provenance & Authenticity · Licensable IP

Prove what's real —
and trace it after it travels.

Fortress binds crop-hardened provenance and per-session watermarks into content, attributes it back to its origin, and guards against false attribution. Sovereign, on-device, no cloud round-trip. Available for licensing or acquisition.

Crop-hardened tiled embedding Per-session leak tracing False-attribution guard · 0/49 null-control On-device · C2PA-adjacent
Two opposite watermarks

The right profile for every use case: RESILIENT or FRAGILE

One engine family, two deliberately opposite behaviours. The SDK hard-wires the profile per facade, so a use case cannot pick the wrong one.

Resilient · survives distribution

Provenance, anti-piracy, leak tracing

The mark is meant to still be there after the content has travelled: compression, re-encode, resize, re-share. Each copy can carry a per-recipient mark, so a leak decodes to its source.

  • 100% recovery vs JPEG and ≤75% downscale, 98% vs noise, 90% vs recolor (source: Fortress benchmark, re-run 2026-09-02, 44 real photos); visibility is judged in ΔE00, not in dB
  • Crop without any side information: content-dependent — photographs 6/6 at crop 10/25/50%, flat graphics fail on a centred crop; 0/6 false hits on unmarked images (lab set of 6 photos — in the shipped SDK, blind centred-crop recovery is still an open gap, see below)
  • Video: H.264 + 10–25% crop, clip-level 8/8 (lab, stealth control 0/8)
  • Boundary, by design: a full generative redraw destroys any blind watermark (seen in our own Gemini redraw tests)
Fragile · breaks on any edit

Authenticity, tamper evidence, science & insurance fraud

The opposite goal: a fragility map that shatters on edits, so a changed region becomes visible and an unaltered capture is provable. Verified by raw sign flips, never by error correction.

  • 0 benign false alarms; 6/6 tamper channels caught, including JPEG q90 and H.264 CRF23
  • Locality: the map shows where the edit happened, not just that one happened
  • The redraw that kills the resilient mark is exactly what the fragile map catches — that boundary is the product
  • Sealed only at live capture (in-app camera). Never applied to an uploaded file: software cannot prove sensor origin, so an upload gets a resilient mark at most

All figures: Research & evidence · reproducible harness under NDA · the fragile verdict is “captured live & unaltered”, never “not AI-generated”.

The differentiator

Provenance that survives the real world

Core capability — measured on 44 real images, real codecs

A tiled, crop-hardened embed carries the proof in every region, so attribution survives the transforms that destroy a naive single-region watermark: JPEG q40–q85 and 50–75% downscale at 100%, noise 98%, recolor 90%, crops 10–50% at 86–92% (source: Fortress benchmark, re-run 2026-09-02, 44 real photos). Origin is bound at the moment of capture, on-device; an adaptive profile self-verifies at embed time and stays fully invisible where the content allows.

Crop-hardened

crops 10–50%: 86–92% recovery (source: Fortress benchmark, re-run 2026-09-02, 44 real photos) — blind centered-crop recovery is a disclosed open gap

Per-session

per-viewer mark for leak tracing; video attribution uses an evidence floor (≥3 frames, ≥60% share)

Guarded

null-control measured: 0/49 false attributions on unmarked assets, 0 hits in 200 unmarked video frames

How we prove it

Measured on the attacks that matter

Fortress is evaluated the way our LLM-trust work is — demonstrated, not asserted. Robustness is evaluated per attack (compression, crop, resize, recapture), imperceptibility measured (PSNR), and — critically — false / phantom attribution is measured against a null control, not assumed away.

  • Recovery reported per attack, including the ones that fail
  • False-attribution rate measured with and without the guard
  • Imperceptibility quantified, not claimed
Honest status — the failures are published with the wins.

The numbers above come from a real, founder-supplied 44-image corpus with real codecs. What does not work, in the same breath: blind centered-crop recovery in the shipped SDK is an open gap — an earlier "crop solved" figure came from a harness that was given the crop factor and key, and we retracted it; a full generative AI redraw destroys any blind watermark (information theory — that boundary is why the fragile tamper layer exists, which flagged AI edits and screenshots in our own Gemini img2img tests); collusion (averaging two copies) defeats per-session tracing — no traitor-tracing codes yet; audio marks survive MP3/AAC/Opus but are audible-ish (SNR 14–17 dB). The reproducible harness ships under NDA. We do not claim "military-grade" or "unbreakable."

Where we focus now

Three problems. Each gets the profile that was measured for it.

01
Fragile + POAW

Insurance fraud & science fraud

Claims photos, lab images and agent decisions that must be provably unaltered — and whose alteration must be visible, with locality.

02
Resilient

Anti-piracy, leak tracing, creator economy, confidential content

Screeners, creator media, high-security and confidential files: every copy carries a per-recipient mark that survives re-share and decodes to the source.

03
Fragile at capture

Authenticity at the moment of capture

Proof-of-real is bound inside our own camera at the shutter — never reconstructed later, never granted to an upload.

All 12 cases, clustered →▶  The 2-minute story Further documented cases: AI-content labelling · newsroom provenance · dataset · text & code
An open invitation

Science fraud: provenance at capture, not in the PDF — we are looking for an instrument maker to test it with

AI can now fabricate micrographs, histopathology and electron-microscope images that reviewers cannot tell from authentic captures. Today provenance is asserted administratively, after capture. There is no cryptographic proof that an image came from a specific instrument, unmodified, at a specific time.

We want to test whether a keyed, fragile integrity map can be bound as early as possible in an instrument’s image path. Today it runs in software on the finished image. A later verifier shows where an image was changed after sealing, for edits of roughly 64 px and larger, including local AI inpainting. Verification can run zero-retention.

  • Feasibility study with a camera or microscope maker on an instrument reference pipeline (not started)
  • 2–3 design-partner labs (academic, pharma QC) to validate the verifier on real instrument output
  • If the study succeeds: a joint licensing term sheet
Honest maturity, stated up front.

The watermark core that proves the concept is working software: in a lab test on 3 images the fragility map raised 0 false alarms and flagged re-save, screenshot, video replay and local inpainting; it misses edits of about 32 px and smaller. It is software only. Anything inside an instrument or sensor is research, with no partner and no hardware yet — that is the co-development scope, not a finished product. We do not claim a mark survives a full generative redraw; for scientific integrity that is the point: fragile by design, so any manipulation is provable. Core methods patent pending (provisionals 03–04/2026). Not a KYC or consumer-identity pitch.

Watch it

Two films, two minutes each

Every number in the films is measured; each film has a scene that says what we do not claim.

Insurance — proof of agent work for insurers

All films, both styles, both languages: /films/

For licensees & acquirers

This IP is available to license or acquire

For camera & capture vendors, DAM and media platforms, creator platforms, and content-authenticity providers who need robust, sovereign provenance with attribution integrity.

ModelWhat it isFit
License + royaltyEmbed Fortress provenance/attribution as an SDK; per-device or per-asset royalty.Capture / media / creator / DAM platforms
Exclusive field licenseUpfront + royalty, exclusive in your field of use.A strategic content-authenticity partner
AcquisitionOne-time asset sale — the provenance IP, SDK, and evaluation harness.Strategic acquirer / corp-dev

Data room under NDA: technical method, patent status, reproducible robustness & attribution benchmarks. SDK evaluation access on request (IP@destill.ai) — Node SDK with 12 per-use-case facades, browser bundle, and the test suite that produced the numbers on this page.

Let's talk

Partner with us on provable authenticity

Licensing, royalty, or acquisition — start with a 20-minute conversation and the evidence pack.

 IP@destill.ai
See NI-Stack (LLM trust layer) →