One brand. One source.
Everything below is the rule.

Section 1

How to use this document

This page is the single source of truth for The AI Lab^™ brand. Where any deck, press release, help article, product screen or partner document disagrees with it, this page wins and the other document is out of date.

It settles four questions:

1. Who we are and what we are for, so that anyone describing us can do it accurately.
2. Which marks are protected, which symbol each one carries today, and how to write them.
3. How the visual system is built, so a TIP surface built by someone else still looks like a TIP surface.
4. Which organisations may use which designations, and what happens when a designation lapses.

The protocol and the brand are licensed separately, and confusing the two is the single most common mistake we see. The specification is open. The name is not.

If you want to use a mark in a way this page does not expressly permit, write to [email protected] with a visual mockup and one paragraph of context. We answer within ten business days.

Section 2

Our mission

The AI Lab Intelligence Unobscured, Inc. is a Delaware C-Corporation founded on January 28, 2026 by Dinesh Mendhe. We build and steward the Trust Identity Protocol^®.

That mission commits us to four things, and we say them out loud because they constrain what we are allowed to build:

The problem we exist to solve

Generative systems made it trivial to fake a person, a photograph, a voice or an entire body of work. Detection does not fix this: a classifier that guesses whether text was machine written will always be one model generation behind, and it accuses innocent people in the meantime. The durable answer is not detection but declaration, signed by a verified human at the moment of publication and checkable by anyone afterwards. That is what TIP^® is.

We use Intelligence, Unobscured. as our line because it names the failure mode we are working against. Intelligence, human or machine, is not the problem. Intelligence passing itself off as something it is not, is.

Section 3

What TIP is, and what it gives you

The Trust Identity Protocol^® is an open, post-quantum standard for two linked things: proving that a real, verified human is behind an identity, and proving where a piece of content came from. It has three layers.

What adopting it gives you

These are the benefits we will state publicly. They are the claims marketing copy, decks and partner material may make. Anything stronger needs sign off.

Why it is built the way it is

• Post-quantum from day one. Signatures use ML-DSA rather than a curve that a future quantum computer retires. A provenance record has to outlive the cryptography that was fashionable when it was written.
• Biometrics never leave the device. Verification uses the platform authenticator, Face ID, Windows Hello or an Android TEE, and a PRF derived key protects the post-quantum key at rest. The network stores no biometric template.
• Federated, not centralised. Anyone can run a full node. The ledger is a DAG replicated across independent operators, so no single company, including us, can switch it off.
• Free where it counts. Verification is free to everyone, always. The licence is free for individuals, nonprofits, journalism, education, government, and businesses under $100K in annual revenue, and converts to Apache 2.0 in 2031.
• Declaration, not detection. The four Origin Codes are declared by the author and bound cryptographically, so accuracy is enforced by reputation and dispute rather than by a classifier guessing.

Section 4

Positioning and boilerplate

Approved descriptions at three lengths. Use them verbatim in press releases, partner pages, conference programmes and speaker bios. If you need a variant, ask rather than improvise.

One line

Short, about 50 words

Full, about 100 words

Facts you may state

Section 5

Voice, tone and writing rules

We write like a standards body that expects to be read by a regulator, a journalist and a developer on the same afternoon. Plain, specific, and willing to state a limitation.

Punctuation, a hard rule

Two characters are banned across everything we publish: prose, interface copy, alt text, aria labels, code comments, JSON content, commit messages and social posts.

The substitution palette is: comma for joined clauses, colon for definitions and list lead ins, vertical bar for meta rows and breadcrumbs, parentheses for a genuine aside, and a full stop when the two halves can stand alone.

Subjects that need care

• Verification Provider economics. The VP economic model is still being designed and is not part of the public launch. Do not describe, imply or estimate a revenue share, commission or per issuance cut. The complete public description of the VP relationship is the annual fee, the independent audit, the quarterly warrant canary, and the jurisdiction declaration.
• Licence tiers. The canonical schedule is nine commercial tiers with a free tier under $100K in annual revenue, published at theailab.org/tip-license. Read that page before writing a threshold or a price. Any earlier five tier schedule, and any $500K free tier ceiling, is retired and must not be republished.
• Trademark symbols. Only two marks carry the registration symbol today. Section 6 is the authority and it changes as certificates issue.
• Scores and disputes. Never present a trust score as a judgement of a person. It is a measure of verification depth and declaration accuracy, and it is appealable.

Section 6

Protected marks and symbol use

The following names and marks are owned by The AI Lab Intelligence Unobscured, Inc. and protected under the U.S. Lanham Act, EU Trade Mark Regulation 2017/1001, the U.K. Trade Marks Act 1994, and analogous laws elsewhere.

Where to put the symbol

• On the first or most prominent appearance of a mark in a document, page or screen. Repeating it in every sentence is not required and reads badly.
• Set as a superscript at the end of the mark, with no space: TIP^®, AI Trust Council^™.
• Not inside a URL, a code identifier, an email address, or a monospaced code sample.
• Not in a page title tag or meta description, where it renders as noise in search results.

Other protected brand elements

• The TIP^® wordmark and the TIP iD icon.
• The four Origin Code letter pairs, OH, AA, AG and MX, as used in connection with TIP^®.
• The visual identity in this document as applied to TIP branded surfaces, including the seven stop brand spectrum.
• The Global Seal of Trust^™ and the AI Trust ID^™ Seal in every colourway, size and tier.
• Domain names containing TIP or TheAILab. See section 25.

Section 7

Writing the TIP name

Section 8

Naming your integration

Implementing TIP^® from the open specification is fully permitted. Naming your implementation after it is not. Choose your own product name, then describe the integration using one of these forms.

Section 9

Logo, wordmark and assets

Use the files below as provided. Do not redraw, recolour, rotate, stretch, animate, outline, add a drop shadow to, or composite the logo with another mark without written permission.

Clear space and minimum size

• Clear space on all four sides equals the cap height of the "T" in the wordmark. Nothing enters that margin, including a partner logo.
• Minimum size is 24px in digital contexts and 0.5in in print. Below that, use the icon alone rather than an illegible lockup.
• On a photograph, place the logo on a plain area or a solid panel. Do not rely on a scrim.

The TIP iD icon in a line of text

• Sits on the same baseline as the surrounding text, at roughly cap height.
• Keeps buffer space of at least half the icon width on every side.
• May be hyperlinked together with the tip:// URI beside it, as one target.

Member, Partner and Verification Provider badges

These are issued by The AI Lab to the holder, not generated by the holder. A badge must be removed within thirty calendar days of the underlying status ending. See sections 20 and 21.

Section 10

Colour

The palette is deliberately small. A surface should be able to name every colour on it, and one accent per component is the rule.

Three colour rules that are not negotiable

1. Every filled brand surface carries white ink. Gold fill or navy fill, the answer is white. Never gold text on navy, never navy text on gold. This applies to buttons, CTAs, pills, chips, active nav states and confirm buttons.
2. Gold is chrome, not text. It belongs on rules, fills, borders, seal artwork and diagram strokes. Gold at 2.50:1 on white cannot carry a sentence. Accent text uses the violet, or the gradient in section 12.
3. Never darken the gold to win a contrast argument. The gold is a fixed brand value. If a contrast number looks awkward, change the ink or the surface, not the brand colour.

Section 11

Typography

Three families, all under permissive open licences, so any integrator can adopt the system without a foundry conversation.

The scale

Rules

• Never ship a heading at weight 600 or 700. The voice is light editorial. A heading that needs more presence gets more size or more tracking, never more weight.
• Display tracking is -0.015em, so the thin letterforms read as deliberate rather than faint. Body tracking is normal.
• Any heading that can wrap uses a line height of at least 1.2. Line height 1.0 is for a single decorative glyph, never for prose.
• Uppercase eyebrows run 11px to 17px at weight 500, tracking 0.14em, in the label ink, preceded by the 1px spectrum tick. An eyebrow labels a field or a kicker. A section heading names a region and stays in sentence case.
• Do not introduce a serif. Cormorant Garamond and Inter are both retired from this system.
• Confirm weight 300 is actually present in the loaded font file. If it is missing the browser synthesises it and the page looks subtly wrong in a way that is hard to diagnose.

Section 12

The spectrum and the accent gradient

Two gradients carry brand signal. They are not interchangeable, and neither is decoration.

The seven stop spectrum, for attention

When a surface needs to say "this one" or "look here", the answer is the spectrum, not a gold tint. Never reorder, recolour or subset the stops.

It appears in exactly two forms.

• The spine. 2px, full height, flush left, over a neutral fill, no border. This marks callouts, notices and active rows. Every emphasised block on this page uses it.
• The outline. The spectrum painted into an existing 1px border with border-box, to mark one card among peers. Never thicken the border to make room, or the card shifts a pixel when it lights up.

The accent gradient, for text

Accent text uses a four stop warm gradient clipped to the glyphs. The amber stop is the point: it is where the old gold lived, so the gradient removes the yellow while keeping the brand warmth. Dropping the warm half does not replace gold, it deletes it.

• box-decoration-break: clone is mandatory. Without it one gradient stretches across every line of a wrapped heading, so the last line lands on the amber stop and renders as the gold you just removed.
• Put the gradient on the element that directly owns the text. A child inherits the transparent fill but not the background image, and renders invisible.
• Never put it in a data record's colour field. That value is consumed downstream as a plain colour, and the result is invisible text.
• Leave single glyphs, arrows, and anything rendered on both light and dark grounds as a solid colour.

Where a gradient cannot go, the solid fallback is two tokens and they are not interchangeable: #7B2FB5 on light grounds and #C9A0F0 on dark. No single violet clears 4.5:1 on both, because the passing luminance ranges do not overlap.

Section 13

Layout, borders, buttons, numerals

Structure

• A soft neutral canvas carrying white cards with 1px borders and generous padding. That is the target whenever the brief is "make it feel modern".
• Base spacing unit 4px, typographic scale 1.125. Section spacing in multiples of 32px, card padding in multiples of 16px.
• Documents and applications are different idioms. A long form document, this page, the licence, the revenue plan, is set flush: square corners, 1px rules that separate rather than enclose, and no fills behind prose. An application is set in cards: radii 14px to 20px, button radii 8px to 14px, neutral fills on a soft canvas. Do not mix them. Rounded tinted panels wrapped around paragraphs are the single clearest sign a page was assembled rather than designed.

Borders and rules

• Left accent bars are 2px maximum. Never 3px, never 4px. A thicker bar reads as a frame around the block rather than a marker beside it.
• Cards, tables and dividers use 1px.
• Selection is an inset ring, not a thicker border, so nothing reflows when a card is chosen.
• Flat fills only on callouts. No gradient washes behind copy.

Buttons

• Filled navy and filled gold are the two primary treatments. Both carry white ink.
• Secondary actions earn their border on hover, not at rest. A bordered button on every row competes with the content.
• Focus ring is a plain 2px gold outline at 2px offset. No companion ring, no halo.

Step numerals

• Banned: a navy disc with a gold numeral, and a gold disc with a navy numeral.
• Invert to a white border and white numeral only when the circle sits on a navy ground.
• Always write a literal fallback inside var(). A border declared against an undeclared custom property is invalid at computed value time, the whole shorthand drops, and the border silently vanishes with no console error.

Mobile

• Touch targets are 44 by 44 CSS pixels minimum.
• CTAs stack full width as pills. No horizontally scrolling chip rails for content.

Section 14

Accessibility

Accessibility is a brand rule here, not a compliance afterthought. A trust layer that cannot be read by everyone is not a trust layer.

• Body text clears 4.5:1. Large text, at 24px or 18.7px bold and above, and non text elements such as rules and icons, clear 3:1.
• Colour never carries meaning alone. Every Origin Code, score tier and status pairs its colour with text, so it survives a screen reader and colour vision differences.
• Every image carries alt text that describes the content, not the file. Decorative elements are marked aria-hidden.
• Focus is always visible: a 2px gold outline at 2px offset, never removed.
• Headings nest in order and are real heading elements, not styled paragraphs.
• Motion respects prefers-reduced-motion.

Section 15

Origin Code badges

The four Origin Codes are the user facing layer of every disclosure obligation TIP^® satisfies, including EU AI Act Article 50. Each is a pill or shield carrying the two letters and the standard colour for that code.

Display rules

• Always the two letter code in capitals. Never "Oh", never "Original-Human" with a hyphen.
• Always the standard colour for that code. Do not invent a new colour, and do not reassign one code's colour to another.
• The chip is a solid fill with white ink, matching the badge artwork. Not a tinted pill with coloured text.
• Two values per code, and they are not interchangeable. The badge fill is the canonical value, used where the text is large. A small chip carrying white ink uses the darkened value, because two of the canonical fills do not clear 4.5:1 at chip text size. See the table below.
• Code and full name appear together on first reference in a document. Later references may use the code alone.
• Origin Code badges may sit alongside the AI Trust ID^™ Seal, the TIP iD icon, a trust score or a timestamp. They never replace the Seal.
• Colour must always be paired with text. See section 14.

OH is darkened by one percent, which is imperceptible. AA is darkened materially, because gold under white ink measures 2.87:1 at the canonical value and is the weakest pairing in the product. AG and MX clear the floor unchanged. Keep the measured ratio in a comment beside any rule that reproduces these.

Section 16

TIP-ID and CTID display

The TIP-ID is the persistent identifier for a verified human or organisation. The CTID is the persistent identifier for a registered piece of content. Both are designed to be printed in bylines, citations and reference lists.

Formats

In a byline

In a citation

Where the badge cannot render

When a CTID is pasted into a social post, an email or a forum that will not render the inline badge, the system appends a two line fallback. This is the canonical form.

Identifier fields in an interface

• A copyable identifier field is neutral everywhere it appears: neutral surface, 1px border, monospaced text at full ink contrast. Never a tinted fill.
• The copy button is a soft neutral pill that turns green on success.
• Hyperlink the URI to the public resolver. The tip:// form may be wrapped as https://tip.theailab.org/r/{ctid} for browser compatibility, or the equivalent resolver of the integrating Verification Provider.

Section 17

The Global Seal of Trust

The Global Seal of Trust^™ is the public certification mark. It has three tiers, each with its own artwork, and a star variant marking founding status.

Sentinel

/seals/sentinel_the_ai_lab_1.webp

Guardian

/seals/guardian_the_ai_lab_1.webp

Sovereign

/seals/sovereign_the_ai_lab_1.webp

• Seal artwork uses its own two values, seal gold #C9A84C and seal navy #0B1629, and does not follow the interface palette.
• The Seal is issued through the Registry. A holder displays the file they were given and never regenerates, recolours or re-tiers it.
• Sentinel, Guardian and Sovereign are the three Seal tiers. They are not licence tiers. The licence schedule is nine commercial tiers published at theailab.org/tip-license, and any earlier five tier naming is retired.
• Seal lettering stays gold. This is one of the places gold is correct as type, because it is artwork rather than interface text.

Section 18

General guidance

Anyone looking at a third party site, app or publication should be able to tell, at a glance, that the third party is not The AI Lab and is not the source of the protocol. Three principles sit behind every specific rule below.

Clarity of source

A reader, regulator or buyer encountering TIP^® anywhere should be able to tell immediately whether they are looking at The AI Lab itself, an accredited Verification Provider, a Member organisation, or an independent implementer working from the open licence.

No suggestion of endorsement

Permitted brand use does not imply endorsement, sponsorship, partnership or any commercial relationship. "Endorsed by TIP", "official TIP partner", "approved by The AI Lab" and "in partnership with TIP" all require a written agreement that exists before the phrase is published.

No appropriation of the marks

Implementing TIP^® from the open specification is fully permitted. Calling your implementation TIP, or any visually or aurally similar variation, is not. See section 8.

Section 19

Apps, products and look and feel

What you may display in your product

• The TIP iD icon, the Origin Code badges, and the AI Trust ID^™ Seal, shown alongside the verification flow they describe.
• A statement that your product implements the protocol, in one of the forms in section 8.
• The colour palette and typography in sections 10 and 11, to harmonise your TIP related surfaces with the rest of the ecosystem.

What you may not

• Use the TIP^® wordmark, or the AI Trust Council^™, AI Trust Registry^™ or Global Seal of Trust^™ marks, as your product name, product logo or visual identity.
• Copy the visual design of theailab.org, vp.theailab.org or any subdomain we operate. Harmonising a badge is not the same as reproducing a page.
• Generate your own Seal artwork. Seals come from the Registry.

Section 20

Verification Provider branding

Verification Providers are organisations accredited by The AI Lab to issue TIP-IDs under TIP Terms of Service Section 6. Accreditation carries specific brand permissions in exchange for specific operational obligations: an annual independent security audit, a quarterly warrant canary, a published jurisdiction declaration, the VP Code of Conduct, and the annual accreditation fee.

Loss of status

A VP that loses accreditation must remove all TIP brand display from its website, marketing materials, product and signage within thirty calendar days. Continued display after loss of status is a trademark violation enforceable under the Lanham Act and analogous laws elsewhere.

Section 21

Member and Partner designations

Badge display

• On your website at a single location, typically the footer, the About page or a dedicated partners page.
• On marketing collateral in the footer or credits area.
• At events, alongside your own signage.

The badge must not be displayed in a way that suggests the holder speaks for TIP^®, operates TIP^®, or publishes these guidelines. It must be removed within thirty calendar days of the status ending.

Section 22

Standards bodies and regulators

TIP^® is designed to be a technical standard input and a compliance mechanism. Co-branding in these contexts is encouraged and follows different rules from commercial co-branding.

Standards bodies

When TIP^® is submitted as a standard contribution to NIST, ISO, IETF, W3C, ENISA, ETSI, BSI or an analogous body, that body's published documents may reference TIP by name and display the wordmark in editorial diagrams and tables. No permission request is needed. We ask for notification, not approval, so the published reference can be cross linked from theailab.org.

Regulators

When a regulator publishes guidance, an advisory or an implementing act referencing TIP^® as a recognised compliance mechanism, the regulator may display the wordmark in that guidance. On request we provide high resolution assets, an executive summary, a technical briefing and a named point of contact.

Section 23

EU AI Act Article 50 co-display

EU AI Act Article 50 transparency obligations became enforceable on August 2, 2026. TIP^® provides open, post-quantum infrastructure that makes Article 50 disclosure routine. See theailab.org/eu-ai-act for the compliance guide. The rules below govern brand co-display in that context.

The Article 50 badge

A platform, publisher or generative AI provider that has implemented Origin Codes on its content surface and subscribed to the published Article 50 Code of Practice may display the badge "EU AI Act Article 50 compliant via TIP" alongside the AI Trust ID^™ Seal on its compliance disclosures, terms of service and corporate site. The badge is issued on request after a brief self attestation. There is no charge for the self attestation.

Co-display with the CE marking

The TIP^® brand may appear in the same regulatory disclosure block as the CE marking, the EU Trust Mark and other EU recognised compliance indicators. It must not be displayed in a way suggesting TIP is itself a CE marking, a notified body designation, or a regulatory authority. TIP is a private sector technical compliance mechanism and the block must make that distinction clear.

Language editions

"Trust Identity Protocol" and "TIP" stay in English in every EU language edition, the way "C2PA" and "GDPR" do. Translators may add a parenthetical translation on first appearance, for example in German "Trust Identity Protocol (Vertrauens-Identitäts-Protokoll, TIP)", but the mark itself is not translated.

European regulator and national authority questions go to [email protected].

Section 24

Community and consortia

Community led sites promoting TIP^® adoption within a region, language community, professional sector or interest area are encouraged.

For a non standard domain pattern or visual approach, write to [email protected] with a mockup and one paragraph of context.

Section 25

Domain usage

Domains we operate

Reserved patterns

A domain in which "tip" or "theailab" or a confusingly similar string is the entire brand portion is reserved to The AI Lab, regardless of top level domain. Existing third party registrations of such domains are addressed under the Uniform Domain-Name Dispute-Resolution Policy and analogous national procedures.

Acceptable patterns

• Path based, the cleanest option: publisher.com/tip, library.university.edu/tip, newsroom.media.com/tip-protocol.
• Subdomain based, where your own name is the parent: tip.publisher.com, tipprotocol.library.org.

Section 26

Misuse and enforcement

We enforce narrowly. The purpose is to protect readers and honest implementers from being misled, not to police every mention of our name. Nominative reference, criticism, comparison, journalism and academic citation need no permission and never will.

What we act on

• A product or company name that incorporates a protected mark.
• A claim of accreditation, membership, partnership, endorsement or certification that does not exist.
• A self generated Seal, badge or designation, or one retained after the underlying status ended.
• A site that reproduces our visual identity closely enough that a reasonable visitor would believe it is operated by The AI Lab.
• Use of ® on a mark that is not registered, including by third parties writing about us.
• A domain registration in which a protected mark is the entire brand portion.

What we do not act on

• Naming TIP^® accurately in an article, review, comparison table or research paper.
• Stating truthfully that a product is built on TIP or is TIP-compatible.
• Criticising the protocol, the Council or the company.
• Forking, modifying or redistributing the reference implementation within the terms of its licence.

Section 27

Attribution, permissions and contact

Required attribution statement

Any document, presentation, marketing material or product interface using a TIP related mark includes this statement at least once, in a footer, credits area or legal notices section. It is reproduced exactly, including the symbols.

Who to write to

Licence of this document

The text of these guidelines is published worldwide under the Creative Commons CC0 1.0 Universal Public Domain Dedication, so any organisation may copy, adapt and fold these rules into its own documentation. The dedication covers the text only. It transfers no rights in the marks the guidelines protect, which remain owned by The AI Lab Intelligence Unobscured, Inc.

Version

Version 2.0, updated August 29, 2026. The current version is always at theailab.org/brand-guidelines. Material changes are announced in advance through the AI Trust Council^™ mailing list and the public archive at theailab.org/insights.