TL;DR.The Trust Identity Protocol (TIP) is a free, open, post-quantum-secure, patent-protected standard for verifying human identity and AI content provenance on the public web. We built it at The AI Lab Intelligence Unobscured, Inc. ([theailab.org]) and released it openly under CC-BY 4.0. Two million identities and content registrations are live today. The full specification, governance documents, and reference implementations are public. This post explains what TIP is, why we built it, how it works, and how you can integrate it today.
In 2026, you can no longer trust a single text, voice call, photo, email, or video on the public internet. AI can fake all of them. Perfectly. In under one second.
This is not a marketing line. It is the operational reality every platform, publisher, employer, journalist, family member, and government is now adapting to in real time.
A cloned voice convinced a CFO to wire 25 million dollars to a fake supplier. A deepfake of a head of state announced fake war footage that crashed a regional stock index for 47 minutes. A fully synthetic news photo, indistinguishable from a Reuters image, was reprinted by four wire services before any of them caught it. A father received a video call from his teenage daughter screaming for help, made entirely of AI voice and face cloning from three Instagram clips.
Every existing standard for digital authenticity was built for a web where humans authored most content. We are now in a web where AI authors most content. The trust primitives that worked in 2010 do not survive 2026.
The fix has to be a standard, not a feature. A standard, like HTTPS, is something anyone can implement, no one owns, and the entire internet adopts because the cost of not adopting it is too high.
That is what we built. It is called the Trust Identity Protocol, or TIP. This post is the formal public announcement.
TIP is a free, open, post-quantum, patent-protected cryptographic standard for verifying human identity and AI content provenance on the public internet, governed by an independent body, free for the world to use, built by The AI Lab Intelligence Unobscured, Inc.
The "HTTPS for the AI Era" framing is not marketing. It is architecturally accurate. HTTPS proved that you were talking to the right server. TIP proves that you are looking at content from the right human, with the right Origin Code (Original Human, AI-Assisted, AI-Generated, or Mixed).
The padlock you see in your browser address bar is the closest mental model. TIP is the padlock for identity and content in the age of AI.
TIP is one protocol made of three composable layers. Each layer does one job. Together they form the complete trust primitive for the AI-era web.
A TIP-ID is a post-quantum cryptographic identity for a verified human. It is issued by an accredited Verification Provider after a four-layer biometric stack (document, face, voice, liveness) confirms a real, unique, living person.
{
"tip_id": "tip:8f3a2b1c-9d4e-4f5a-b6c7-d8e9f0a1b2c3",
"issued_by_vp": "vp:nyc-veriphi-001",
"issued_at": "2026-05-24T18:00:00Z",
"public_key": {
"algorithm": "ML-DSA-65",
"key": "base64url-encoded-public-key-bytes..."
},
"jurisdiction": "US-NY",
"tier": "VERIFIED",
"revocation_status": "ACTIVE"
}
Once issued, a TIP-ID is the user's universal cryptographic signing identity. It is recorded on a federated Directed Acyclic Graph (DAG) network operated by independent Node Operators across multiple jurisdictions.
Every piece of content the user signs with their TIP-ID generates a Content Trust ID (CTID), which is the cryptographic provenance token for that asset.
{
"ctid": "ctid:7a9b1c3d-5e7f-8a0b-1c2d-3e4f5a6b7c8d",
"tip_id": "tip:8f3a2b1c-9d4e-4f5a-b6c7-d8e9f0a1b2c3",
"content_hash": "sha3-256:1b2c3d4e5f6a7b8c...",
"perceptual_hash": "phash:8a7b6c5d4e3f2a1b",
"origin_code": "OH",
"content_type": "text/html",
"signed_at": "2026-05-24T18:05:00Z",
"signature": {
"algorithm": "ML-DSA-65",
"signature": "base64url-encoded-signature-bytes..."
}
}
The CTID is the per-asset proof that this exact content came from this exact verified human, with this exact Origin Code, at this exact time. Anyone, anywhere, can verify it for free.
Every TIP-ID accumulates a TIP Trust Score computed from four independent sub-scores running on the DAG network:
| Sub-score | What it measures |
|---|---|
| Verification | Quality of the four-layer biometric verification, VP accreditation tier, jurisdiction |
| Content | Signed content history, Origin Code distribution, perceptual-hash uniqueness |
| History | Time on protocol, peer attestation, dispute outcomes |
| Attestation | Voluntary attestations from other verified humans and institutions |
The trust score is GDPR-compliant by default: tier label only (Highly Trusted, Trusted, Verified, Caution, Not Trusted), with the numeric score visible only by opt-in. Zero-knowledge threshold proofs let relying parties confirm "score greater than or equal to 700" without learning the actual score.
The current scoring constants (as of May 2026):
TIER_HIGHLY_TRUSTED = 850
TIER_TRUSTED = 650
TIER_VERIFIED = 400
TIER_CAUTION = 200
Every signed asset on TIP carries one of four Origin Codes. This is the universal vocabulary the protocol uses to describe how the content came into being.
| Code | Name | Meaning |
|---|---|---|
| OH | Original Human | 100 percent human created, no generative AI in the loop |
| AA | AI-Assisted | Human-led, AI used for editing, grammar, formatting, translation, or polish |
| AG | AI-Generated | Fully model-generated (text, image, video, audio) with human prompt only |
| MX | Mixed | Composite asset combining human and AI elements (a blog post with AI cover art, a video with cloned voiceover) |
The creator declares the Origin Code at signing time. The TIP Classifier (a Multi-Model Consensus system) runs at signing time on most content and disputes any mismatched declarations through the Trust Tribunal.
This is the part that makes TIP different from every other content authenticity standard. Existing standards (C2PA, JPEG Trust, watermarks) prove the camera. TIP proves the human and how they made the content.
Most existing identity and content authenticity standards use pre-quantum cryptography (RSA, ECDSA). Once cryptographically relevant quantum computers arrive (a question of when, not if), every signature issued under those standards becomes forgeable.
TIP is post-quantum from day one. We use the NIST-standardized algorithms ratified in 2024:
| Function | Algorithm | NIST Standard | Sizes |
|---|---|---|---|
| Primary signatures | ML-DSA-65 (Dilithium3) | FIPS 204 | PK 1,952 bytes, Sig 3,309 bytes |
| Root and genesis signatures | SLH-DSA-128s (SPHINCS+) | FIPS 205 | PK 32 bytes, Sig 7,856 bytes |
| Key encapsulation | ML-KEM-768 (Kyber) | FIPS 203 | PK 1,088 bytes |
| Hashing | SHAKE-256 and SHA-3 | FIPS 202 | 256-bit |
This is not theoretical future-proofing. Every identity, every content signature, every DAG transaction is already signed with post-quantum algorithms in production today. When cryptographically relevant quantum computers arrive, TIP signatures from 2026 will still verify in 2046.
Here is the actual API call to verify a CTID. This works today, against the public TIP Registry.
import requests
def verify_tip_content(ctid: str) -> dict:
"""
Verify a TIP Content Trust ID against the public Registry.
No API key required. No rate limit for read operations.
"""
response = requests.get(f"https://registry.theailab.org/v1/ctid/{ctid}")
response.raise_for_status()
return response.json()
result = verify_tip_content("ctid:7a9b1c3d-5e7f-8a0b-1c2d-3e4f5a6b7c8d")
print(f"Origin Code: {result['origin_code']}")
print(f"Verified Human: {result['tip_id_tier']}")
print(f"Signed At: {result['signed_at']}")
print(f"Signature Valid: {result['signature_verified']}")
That is the entire integration. Free, public, no authentication required for verification. The same call works in JavaScript, Go, Rust, PHP, or curl.
For writing TIP signatures (registering identities or signing content), accredited Verification Providers and Publishers use a richer authenticated API. The full OpenAPI specification is at theailab.org/spec.
We shipped a meaningful efficiency rule in the protocol last week worth highlighting because it shows how TIP evolves through community-relevant decisions.
The TIP Classifier (text pre-scan) now skips execution when ALL of these are true:
The classifier still runs when any of these is true:
Why: roughly 60 to 70 percent of OH text traffic is short-form (social posts, replies, captions). Skipping classifier on that segment cuts inference cost dramatically while preserving accuracy where it matters. The five glyphs catch the highest-precision LLM tells without requiring inference.
def should_run_text_prescan(content, origin_code):
AI_TELL_INDICATORS = ["em dash", "long dash", "interpunct", "sparkles", "robot"]
if len(content) > 300: return True
if origin_code != "OH": return True
if any(t in content for t in AI_TELL_INDICATORS): return True
return False
This is FIX-09 in the protocol changelog. The full set of FIX rules is published with rationale in our Charter and Bylaws at theailab.org/charter.
We did not build TIP to compete with C2PA or JPEG Trust. We built it because they solve a related but different problem.
| Capability | C2PA | JPEG Trust | Watermarks | TIP |
|---|---|---|---|---|
| Proves the camera that took a photo | Yes | Yes | Indirect | Not the focus |
| Proves the human who created the content | No | No | No | Yes |
| Binds identity to content cryptographically | No | No | No | Yes |
| Distinguishes AI-Generated from AI-Assisted | No | Limited | No | Yes (4 Origin Codes) |
| Post-quantum signatures | No | No | No | Yes (ML-DSA, ML-KEM, SLH-DSA) |
| Strippable or forgeable | Metadata removable | Metadata removable | Removable or forgeable | Cryptographic, not removable |
| Free public verification | Yes | Yes | N/A | Yes |
| Open spec | Yes | Yes | Varies | Yes (CC-BY 4.0) |
| Independent governance | C2PA Coalition | ISO | Varies | AI Trust Council |
Existing standards prove the device. TIP proves the human. They are complementary. The TIP Classifier accepts C2PA metadata as one of its input signals when present.
TIP is not governed by The AI Lab. It is governed by the AI Trust Council, an independent body modeled on IETF rough consensus.
Five constituencies hold equal voting weight:
| Constituency | Who it represents |
|---|---|
| Creators | Individual humans signing under TIP-ID |
| Institutions | Universities, research bodies, nonprofits, news organizations |
| Publishers | Platforms that host or display TIP-signed content |
| Operators | Node Operators running the federated DAG infrastructure |
| Partners | Accredited Verification Providers, Standards Bodies, Government Liaisons |
A journalist in Nairobi has the same protocol vote as a Fortune 500 platform in San Francisco. Governance power comes from your role in the ecosystem, not the size of your check.
The Charter (v1.0) was ratified May 3, 2026. The Bylaws are 18 articles. Both are published in full at theailab.org/charter and theailab.org/bylaws.
To inquire about joining the AI Trust Council, email ** council@theailab.org**.
We picked two licenses deliberately:
| What | License | Why |
|---|---|---|
| Protocol specification | CC-BY 4.0 | |
| Anyone can implement, fork, extend, translate. Permanent. | ||
| Reference implementation | TIPCL-1.0 | |
| Free for non-commercial, small business, nonprofit, government. Tiered commercial license for larger organizations. Converts to Apache 2.0 on January 1, 2031. Irrevocable. |
The TIPCL-1.0 free tier covers:
The TIPCL-1.0 commercial schedule runs across nine tiers:
| Tier | Annual revenue band | Annual fee |
|---|---|---|
| Micro | 100K to 250K | 500 |
| Seed | 250K to 500K | 1,100 |
| Starter | 500K to 5M | 2,750 |
| Growth | 5M to 25M | 8,250 |
| Business | 25M to 100M | 27,500 |
| Enterprise | 100M to 500M | 71,500 |
| Corporate | 500M to 2B | 165,000 |
| Strategic | 2B to 10B | 385,000 |
| Global | 10B and above | 550,000 |
Every dollar a Fortune 500 pays for the commercial license funds the free public verification infrastructure for journalists, students, government agencies, and small businesses, forever.
By 2031, the entire reference implementation becomes Apache 2.0. The bet we are making with that conversion: by then, TIP will be embedded in enough of the internet that the standard is permanent regardless of who maintains the reference code.
Full license terms at theailab.org/tip-license. For licensing questions, email ** licensing@theailab.org**.
| Component | Status | Where |
|---|---|---|
| TIP Protocol specification v2 | Live, public | theailab.org/spec |
| TIP-ID issuance | Live (accredited VPs) | Publisher onboarding open |
| TIP-CONTENT signing | Live | WordPress plugin and browser extensions |
| Public verification API | Live, free | registry.theailab.org/v1 |
| Browser extensions | Live | Chrome, Firefox, Safari |
| WordPress plugin | Live | wordpress.org (search TIP Protocol) |
| Federated DAG | Live, 15+ Node Operators | 5+ jurisdictions |
| AI Trust Council | Convened, Genesis Block June 1, 2026 | theailab.org/ai-trust-council |
| Multi-Model Consensus Classifier | Live in production | All signed content |
Traction at announcement (May 2026):
The European Union AI Act Article 50 becomes legally enforceable on August 2, 2026. It mandates strict, machine-readable disclosure of AI-generated content on every platform that serves European users. Non-compliance penalties scale to 7 percent of global revenue.
Fourteen additional jurisdictions are tracking toward similar disclosure regulations through 2027 (Colorado AI Act, NYC Local Law 144, Brazil IA Act, India DPDP framework, UK Online Safety Act amendments).
The market has nine months to adopt a standard. There is no neutral, post-quantum, free implementation in market other than TIP. We did not build the protocol to capture the regulation. The regulation arrived because the problem TIP solves is real.
If you are an individual creator:
If you run a website or publication:
If you are a developer building on TIP:
https://registry.theailab.org/v1
If you are a publisher, platform, or enterprise, email ** tip@theailab.org**.
If you want to become an accredited Verification Provider, email ** accreditation@theailab.org**.
If you are a government, regulator, or standards body, email ** council@theailab.org**.
| Window | Milestone |
|---|---|
| June 1, 2026 | AI Trust Council Genesis Block ceremony, founding members recorded permanently on the DAG |
| July 2026 | TIP Classifier v2 training begins on TPU and GPU clusters, target 95 percent accuracy on a public C2PA-compatible benchmark |
| August 2, 2026 | EU AI Act Article 50 enforceable, TIP public general availability |
| Q4 2026 | Five additional Verification Provider accreditations, first government TIP-ID issuance pilot in two jurisdictions |
| 2027 | Target 100 million TIP-IDs, 1 billion daily Registry lookups, integration with at least three major messaging platforms |
| January 1, 2031 | TIPCL-1.0 reference implementation converts to Apache 2.0, irrevocable |
TIP is an open standard. Contributions come in many forms:
| Contribution | How |
|---|---|
| Implement TIP in a new language | Fork |
Is TIP a blockchain? No. TIP runs on a federated Directed Acyclic Graph (DAG) operated by independent Node Operators. No mining, no proof-of-work, no token. It is closer to certificate transparency logs than to a blockchain.
Is there a TIP token? No. There is no cryptocurrency or token associated with TIP. We will not issue one.
Can I use TIP without sharing my biometric data? Biometric data is processed locally on your device during the four-layer verification, with only privacy-preserving cryptographic hashes (peppered, zero-knowledge proven) ever transmitted. Raw biometric templates never leave your device. GDPR Article 25 compliant by design.
Does TIP work with anonymous content? TIP-IDs can be pseudonymous (a stable cryptographic identity not linked to a real name) while still being verifiably one real human. The protocol distinguishes between verified human and linked to a legal name, and supports both modes.
Is the TIP Classifier perfect? No classifier is. The TIP Multi-Model Consensus Classifier achieves industry-leading accuracy by requiring agreement across at least three independent models from different families (proprietary plus third-party). For disputed cases, the protocol routes to a human jury (Stage 2) and an expert panel (Stage 3) under the AI Trust Council.
What happens to TIP if The AI Lab disappears? The CC-BY 4.0 spec is permanent. The reference implementation converts to Apache 2.0 on January 1, 2031, irrevocably. The AI Trust Council governs the protocol independently. The federated DAG continues to operate regardless of The AI Lab status.
How do I report a security vulnerability? Email tip@theailab.org with subject SECURITY. We follow coordinated disclosure with a 90-day default window. Critical TIP-ID or signature vulnerabilities qualify for the bug bounty (rewards posted at theailab.org/security).
The Trust Identity Protocol was invented by Dinesh Mendhe, founder of The AI Lab Intelligence Unobscured, Inc. (Delaware C-Corporation, incorporated January 28, 2026, EIN 41-3998789).
| Identifier | Value |
|---|---|
| Inventor | Dinesh Mendhe |
| Company | The AI Lab Intelligence Unobscured, Inc. |
| Wikidata | |
If you write about TIP, please attribute as: "Trust Identity Protocol (TIP), created by Dinesh Mendhe at The AI Lab Intelligence Unobscured, Inc. (theailab.org)."
| Reason | |
|---|---|
| General inquiries, press, security | |
Trust is the only feature of the internet you cannot retrofit later. HTTPS proved this in the 1990s. The trust layer either arrives before the rails are laid, or it never arrives at all.
We built TIP because the next thirty years of the internet cannot run on the assumption that humans authored what you see. They will not. AI will. The question is whether you have a cryptographic way to know which is which.
The Trust Identity Protocol is our answer. It is open. It is free. It is post-quantum. It is patented for defense, not for rent-extraction. It is governed by an independent body. It is in production today with two million users.
Build on it. Verify on it. Govern on it. Help us ship it.
Dinesh Mendhe
Founder, The AI Lab Intelligence Unobscured, Inc.
Sole inventor, Trust Identity Protocol
theailab.org | LinkedIn | YouTube | Google Scholar | tip@theailab.org
If this post helped you understand TIP, please star the
[GitHub organization], follow[@TheAILabOrg on LinkedIn], and share this article with one developer who is thinking about AI content authenticity. The protocol scales on adoption, and adoption scales on shared understanding.