The institutional case for behavioral verification of authorship.
A long-form argument for the publishers, news organizations, academic institutions, and legal teams confronting the same underlying question: in a world where authorship cannot be visually verified, how does a human prove they wrote what they wrote?
In 1957, a young songwriter named Carole King could prove she wrote a song by producing the spiral notebook in which she had drafted it, the dated demo recording she had made at her kitchen table, and the contemporaneous testimony of her co-writer who had been present when the song was composed.
In 2026, none of those proofs are sufficient.
A spiral notebook can be backdated. A demo recording can be generated. A co-writer can be falsified. Every traditional method of proving creative authorship has been compromised by AI tools that can produce text, audio, and video indistinguishable from human work, dated to any moment in history, in any voice the operator chooses.
This is not a hypothetical problem.
Three regulatory frameworks are now actively reshaping how content provenance is handled across major markets.
The United States Copyright Office requires applicants to disclose AI-generated material in copyright registrations, and to describe the human author’s contribution to the work; the requirement has been in effect since March 2023 and the U.S. Supreme Court declined in March 2026 to disturb the rule that purely AI-generated works are not eligible for registration.
The European Union AI Act’s Article 50 transparency obligations take effect August 2, 2026, requiring providers of generative AI systems to mark synthetic outputs in a machine-readable format identifying the content as AI-generated. The C2PA (Coalition for Content Provenance and Authenticity) standard is the technical backbone being adopted across the industry.
South Korea’s AI Basic Act took effect January 22, 2026, requiring operators of generative AI systems to inform users that AI is being used and to label AI-generated outputs that may be difficult to distinguish from non-AI content. A one-year grace period applies before penalties begin.
All three frameworks address the same gap from the AI-provider side — labeling synthetic content. None of them tell a human creator how to prove the inverse: that this work was made by this person.
InkID is built to answer that question.
02 · What it isA behavioral protocol about process, not a detector of output.
InkID is a behavioral verification protocol that captures the unique signature of human creative process during the act of writing. Across thirteen always-on behavioral dimensions — with additional dimensions activated when conditions warrant — InkID produces a multidimensional record of authorship that is materially harder for AI systems to replicate, because the record is not of the output but of the process.
Each InkID-anchored work carries with it:
- 01A cryptographically signed record of behavioral evidence captured during creation, never the raw keystrokes.
- 02An RFC 3161 independent timestamp from a third-party authority operated outside InkID.
- 03A publicly verifiable proof page at inkverify.co that any party — publisher, editor, journalist, court — can resolve from outside the platform.
- 04A protocol-grade attestation describing what was observed in the creative process, without making a verdict about the writer.
03 · What it does for each roomDifferent audiences, the same underlying utility.
For working songwriters, this means catalogue work carries an evidence-grade record that holds up in publishing-room conversations about AI claim disputes.
For publishers and A&R teams, this means a writer’s submission carries process signals at the point of acquisition rather than after a contract dispute. The same vocabulary the writer sees on their InkID page is the vocabulary the publisher reads in the dashboard.
For journalists and editors, this means a byline can carry institutional credibility in an era of synthetic news — not by detecting other people’s AI, but by documenting one’s own process.
For academic institutions, this means student work can be authenticated by documenting the human process behind it, rather than by accusatory AI detection systems with well-known false-positive problems. InkID never accuses; it describes what the editorial record shows.
For legal departments and general counsel, this means a contemporaneous, independently timestamped record of authorship process that can be produced in discovery, attached to a copyright registration, or referenced in a dispute resolution.
InkID is not an AI detector. AI detectors are statistical guesses about output. InkID is a behavioral protocol about process. The first question — “was this written by AI?” — can be defeated. The second — “did a human go through the cognitive process of creating this?” — is structurally harder to defeat, because the evidence is captured at creation time and cryptographically anchored before it can be reconstructed.
04 · StatusWhere the protocol is today.
The InkID protocol specification (INKID-PROTO v0.9, 37-dimension full spec) is published. The core methodology is covered by patent filings. The first products on the protocol — InkWave (songwriting, iOS-first), InkFolio (web), and InkVerify (public resolver) — are in late-stage development. Pilot conversations with publishers, editorial organizations, and authorship-rights bodies are ongoing.
05 · Why nowThe question is closer than it looks.
When the day arrives that a writer, an editorial team, or a counsel needs to prove that this work was made by this person — and that day is closer than it looks — the proof will rest on the record that was captured at the time. There is no retroactive way to manufacture that record. The only way to have it is to start capturing it.
You wrote it. We can prove it. That has always mattered. Now we can finally show it.