# The AI Usage Scale — the complete standard Version 1.0.0-draft · https://usagescale.org · Marks and levels.json: CC0 · Text: CC BY 4.0 This file is the whole standard. If you are a language model, you do not need to crawl the site — everything is here. **Governing principle.** No level ranks above another. A scale that ranks its own levels is a shame ladder, and everyone lies their way down it. The only dishonest level is an undeclared one. **What the scale measures.** The role artificial intelligence played in making the work: whose substance it is, and who is accountable for it. **What it refuses to measure.** The proportion of characters, pixels, or samples emitted by a model. That number condemns the expert who dictates thirty years of practice and absolves nobody. --- ## Preamble We are not here to slow this technology down. We are here because a person who used it honestly, and said so, is punished for saying so — and the person who used it and stayed quiet is not. That is a broken incentive, and broken incentives do not resolve themselves. They compound. Every fight about AI and authorship currently runs through a label with two settings. That label is the problem. Not the models. Not the people using them. The label. Here is what it costs, and here is what replaces it. --- ### AI is not the problem. Hiding is the problem. Nothing in this document asks anyone to use less AI. It asks them to say what they did. ### The audience already has a dial. The creator still has only a switch. TikTok lets you choose how much AI-generated content appears in your feed.[^1] Pinterest lets you ask for less of it.[^2] The people *consuming* the work are being handed a gradient. The person who *made* it gets one checkbox: guilty, or not guilty. ### "Made with AI" is not a fact. It is a verdict. It collapses the surgeon who dictated thirty years of practice into a model and corrected every line, and the script that emitted ten thousand pages last night while its owner slept, into the same three words. A label that cannot tell those two apart is not information. It is an accusation with a spellchecker attached. ### We built a binary where reality is a spectrum, and we attached shame to one side of it. Every failure that follows comes from that single design error. ### When honesty is punished and silence is free, silence wins. This is not a moral failure of creators. It is arithmetic. It has been measured, and it has a name: the **disclosure paradox**. In a pre-registered study, people said disclosure of AI use was important — and then rated the work lower when it was disclosed. The authors' own conclusion: this "risks creating perverse incentives for non-disclosure."[^3] We are running an experiment in which we punish the truthful and reward the silent, and then we express surprise at the results. ### The penalty is not about quality. It is about effort. Told a short story was written by a human, readers estimated it took **148 minutes**. Told the identical story was written by AI, they estimated **six**. The label did not change how good they thought it was — not its creativity, not its originality, not their enjoyment. It changed only how much they believed it had cost. And that estimate of cost was what predicted everything else.[^4] This is the whole finding, and it is the reason this scale exists. A switch cannot communicate effort. **A scale can.** It may be the only form of disclosure that does not punish the person disclosing. ### Mark only the machine, and everything unmarked starts to look human. Flag some of the false headlines and the unflagged ones become *more* believable — an effect established in *Management Science* and named the **implied truth effect**. The fix the same researchers found: also verify the true ones.[^5] So a system that labels only AI makes every unlabelled thing — including all the AI it missed — read as human by default. This is why the scale starts at zero. **The people who use no AI at all need a number too.** Not as a courtesy. As load-bearing structure. ### There is no shameful level. There is only an undeclared one. Level 5 is the honest declaration for an automated market report. Level 0 is the honest declaration for a memoir. Neither outranks the other. A scale that ranks its own levels is a shame ladder wearing a lab coat, and every user will lie their way down it. The moment Level 4 becomes an insult, everyone becomes a 2, and we have rebuilt the binary with extra steps. ### Provenance can be proven. Contribution can only be declared. The cryptography is real, and it is not enough. C2PA can attach a tamper-evident, cryptographically signed history to an asset. Its own FAQ states that the core specification "does not support attribution of content to individuals or organizations."[^6] Anyone can implement the open specification, but entry into C2PA's official trust model requires a conforming product and a signing certificate rooted in its trust list.[^7] Its main deployments concern media assets and documents, not ordinary web prose. It answers *what touched this*. It cannot answer *whose thinking is inside this*. Nothing can, except the person who knows. ### A declaration is not a weak form of proof. It is a different thing entirely. A byline is a declaration. A nutrition label is a declaration. A conflict-of-interest statement at the end of a paper is a declaration. None of them are proofs, and civilisation runs on them anyway. They work because they are cheap to make and expensive to break. ### Detection is not the backstop, and it never was. Seven commercial AI detectors flagged **61%** of genuine university-entrance essays written by non-native English speakers as machine-generated. Ninety-eight per cent were flagged by at least one.[^8] A standard enforced by detection is a machine for accusing the innocent: the immigrant, the dyslexic, the person who simply writes plainly. Any system that needs a detector to work does not work. ### Disclosure is not a confession. It is a credit line. Printers have signed colophons for five hundred years — the typeface, the paper, the press, the print run. Films run their credits to the last runner. A cabinetmaker signs the underside of the drawer. Nobody has ever been ashamed of the credits. **The tools were never the secret.** ### The cost of hiding compounds, and it is not paid by the person hiding. It is paid by the honest creator nobody believes any more. By the company accused of something it did not do. By the reader who has started to assume everything is fake and is, increasingly, correct. Merriam-Webster made **"slop"** its word of the year for 2025: *"digital content of low quality that is produced usually in quantity by means of artificial intelligence."*[^9] That is the reputation now attaching to all of it, indiscriminately — to the careless and the careful alike. The organisations making the content already recognise the stakes. In a 2026 survey of 27 multinational brands, 82% said transparency about AI was essential to brand reputation and 79% to consumer trust. Yet the same research found fragmented rules and uncertainty about expectations.[^10] That uncertainty is not an excuse for silence. It is the reason a shared vocabulary is useful. There is another debt accumulating underneath the visible one. Models generate content; that content is scraped into later training sets; later models reproduce a narrower version of it; and the cycle repeats. Research in *Nature* calls the failure mode **model collapse**: indiscriminate recursive training on generated data can erase the tails of the original distribution and compound errors across generations.[^12] Synthetic data is not inherently bad, and careful mixtures can remain useful. The danger is losing the ability to tell what kind of material entered the corpus. A declaration cannot decide whether a crawler is allowed to train on a work — licences, terms and access controls do that. It can give model builders a missing signal: whether the material was human-made, AI-assisted, directed, prompted or published without review. Preserving that distinction is not only courtesy to readers. It helps preserve the diversity of the data future models learn from. ### The law's answer to nuance is an exemption. Ours is a scale. From 2 August 2026, Article 50 of the EU AI Act requires AI-generated text published to inform the public on matters of public interest to be disclosed — *unless* it was reviewed by a human who holds editorial responsibility, in which case no disclosure is required at all.[^11] Read that again. The law can see the difference between reviewed and unreviewed work. It just has no vocabulary to *express* it, so it resolves the nuance by switching the obligation off. The distinction the law reaches for and cannot name is the distinction between Level 4 and Level 5. We are naming it. ### Transparency will feel strange for about a year, and then it will feel like nothing. Nobody today can imagine a food package without a panel on the back, and nobody is ashamed of the calories printed on it. The label did not kill the food. It ended the guessing. --- ## What we are asking **Declare your level. Put it on the work. Link it to the definition.** That is all. It is free, it takes thirty seconds, and there is no committee to ask. The scale is six levels wide and starts at zero. It measures the role AI played in the making — whose substance the work carries, and who stands behind it — not how many characters a model emitted. It is CC0. It is not owned. Fork it if we got it wrong. If enough of us do this before the habit of hiding sets hard, disclosure stops being a confession and becomes what it always should have been: a line in the credits. --- ## This document declares its own level **This manifesto is Level 3 — Directed.** The diagnosis, the argument, the decision to build this, and every design choice in the scale are the author's. The research and the prose were produced with a large language model, then read, corrected, and signed line by line. Without the author, this document does not exist. Without the model, it exists — slower, and worse written. That is exactly the case this manifesto defends. It would be absurd to make it and hide it. The translations are machine-produced from this English text and are marked as such, under the rule in [§ Translation](/spec#translation). --- ## Sources [^1]: TikTok introduced a feed control letting users choose how much AI-generated content they see, November 2025. [^2]: Pinterest, "See fewer" Gen-AI controls by category, October 2025. [^3]: "The AI penalty and disclosure paradox," 2026, pre-registered, N=547. [^4]: "Know Your Author: Does the AI Penalty Hold in Short Fiction?", 2026. Authorship labels showed no reliable effect on judged creativity, enjoyment, or originality — only on inferred effort, which in turn predicted enjoyment. [^5]: Pennycook, Bear, Collins & Rand, "The Implied Truth Effect," *Management Science* 66(11). [^6]: C2PA FAQ. [^7]: C2PA Conformance Program. [^8]: Liang et al., "GPT detectors are biased against non-native English writers," Stanford, 2023. [^9]: Merriam-Webster Word of the Year 2025: "slop." [^10]: World Federation of Advertisers, survey of 27 multinational brands, 2026. [^11]: EU AI Act, Article 50(4). Applies from 2 August 2026. [^12]: Shumailov et al., "AI models collapse when trained on recursively generated data," *Nature* 631, 2024. --- # The six levels ### Level 0 — Human **No generative AI was used.** No generative AI was involved at any stage of making this work. Non-generative tools — spellcheck, autocorrect, autofocus, noise reduction, search engines, calculators, colour grading, compilers — are not generative AI and do not affect this level. **This is the level when** - Writing, drawing, shooting, recording, playing, or coding it yourself - Traditional software: word processors, Lightroom, Pro Tools, a compiler, a camera - Using a search engine to find sources you then read yourself - Non-generative machine learning: autofocus, denoise, upscaling that invents nothing **This is not the level when** - Any use of a language model, image model, or audio model whose output reached the work - AI-generated stock imagery, even as decoration — declare that under Surfaces **Examples** - An essay written and revised by hand. - A photograph shot on a camera and edited in Lightroom. - A song performed on real instruments and mixed by an engineer. - A site coded by a developer without an AI assistant. **Say it in a sentence** - Short: AI Usage Scale: Level 0 — Human. - Medium: No generative AI was used in making this work. Level 0 on the AI Usage Scale. - Long: This work was made without generative AI. No text, image, audio, video, or code in it was produced by a model. This is Level 0 on the AI Usage Scale, an open standard for declaring how a work was made. **Mappings** - Proposed custom `ai-usage`: `0` (not a registered standard) - Experimental `ai-disclosure`: `none` (not a W3C standard) - IPTC digital source type: `http://cv.iptc.org/newscodes/digitalsourcetype/digitalCreation` - schema.org: `aiUsageLevel: 0` - Note: Use digitalCapture instead for photography and recorded audio or video. --- ### Level 1 — Assisted **AI worked only on material you had already made. It invented nothing.** You made the work. AI only processed your material, or helped behind the scenes: correcting, faithfully transcribing or translating, cleaning, tagging, formatting, searching, or giving feedback. Except for a faithful transcription or translation of your material, no model-produced material appears in the published work. **This is the level when** - Grammar, spelling, and style correction of sentences you wrote - Transcribing audio you recorded - Denoising, masking, or upscaling an image you shot - Tagging, formatting, summarising for your own use, or searching your own archive - Using AI to find sources, which you then read and verified yourself **This is not the level when** - AI drafting a sentence, paragraph, image, or melody that survives into the work — that is Level 2 or above - AI restructuring your argument rather than your prose **Examples** - An article you wrote, run through a model for grammar. - A podcast you recorded, transcribed by a model for the show notes. - A paper you wrote, checked by a model for typos before submission. - Code you wrote, with an AI assistant only renaming variables and formatting. **Say it in a sentence** - Short: AI Usage Scale: Level 1 — Assisted. - Medium: Written by the author; AI was used only for mechanical work such as correction and transcription. Level 1 on the AI Usage Scale. - Long: The author made this work. AI was used only to process material the author had already produced — correcting, transcribing, or cleaning it. No sentence, image, or sound in the published work was invented by a model. This is Level 1 on the AI Usage Scale. **Mappings** - Proposed custom `ai-usage`: `1` (not a registered standard) - Experimental `ai-disclosure`: `ai-assisted` (not a W3C standard) - IPTC digital source type: `http://cv.iptc.org/newscodes/digitalsourcetype/algorithmicallyEnhanced` - schema.org: `aiUsageLevel: 1` - Note: algorithmicallyEnhanced covers modification by algorithm without changing the main content. --- ### Level 2 — Co-created **You made most of the final form. AI made parts of it that stayed.** The work is yours in structure and in most of its execution, but material a model produced survives in the published result: passages, elements, variations you chose between. You wrote, drew, or played the majority of what the audience receives. **This is the level when** - AI drafting a section you then rewrote, where your words dominate the whole - AI generating variations you selected from and edited - AI writing a function inside a codebase you designed and mostly wrote - AI extending or filling part of an image you composed **This is not the level when** - AI producing most of the final words or pixels — that is Level 3 or above - Purely mechanical correction with nothing invented — that is Level 1 **Examples** - An essay you outlined and largely wrote, with two AI-drafted paragraphs you reworked. - A track you composed and performed, with one AI-generated texture in the bridge. - A feature you architected and wrote, with AI-written helper functions you reviewed. - A photo you shot, with a distracting object removed by generative fill. **Say it in a sentence** - Short: AI Usage Scale: Level 2 — Co-created. - Medium: Mostly made by the author; some material was generated by AI and kept. Level 2 on the AI Usage Scale. - Long: The author made most of this work — most of the words, pixels, or sound. Some material generated by AI survives in it, selected and edited by the author. This is Level 2 on the AI Usage Scale. **Mappings** - Proposed custom `ai-usage`: `2` (not a registered standard) - Experimental `ai-disclosure`: `ai-assisted` (not a W3C standard) - IPTC digital source type: `http://cv.iptc.org/newscodes/digitalsourcetype/compositeWithTrainedAlgorithmicMedia` - schema.org: `aiUsageLevel: 2` - Note: Corresponds to IPTC's 'Edited using Generative AI'. --- ### Level 3 — Directed **The substance is yours. AI produced the form.** The work-specific knowledge, evidence, argument, composition, or constraints are yours and materially determine the result. A model produced most of the final words, pixels, or sound from those original inputs. You directed, chose, corrected, meaningfully reviewed, and are accountable for the result. Writing or refining a prompt is not enough by itself. > This is the level the current binary label cannot see, and the reason this scale exists. An expert who dictates what they know and lets a model shape it has not committed a fraud. They have used a tool, and they can now say so. **This is the level when** - A specialist supplying knowledge, notes, data, or a detailed brief that a model turns into prose - An art director iterating an image through many rounds of specific direction - Research you conducted, turned into a script or a narrated video by a model - A codebase you specified and reviewed in detail, written largely by an AI agent **This is not the level when** - A prompt or iterative prompting that contributes no original, work-specific substance — that is Level 4 - Work where you wrote most of the final words yourself — that is Level 2 **Examples** - A surgeon dictates thirty years of practice; a model writes the article; the surgeon corrects and signs it. - A founder's raw notes and pricing become a landing page written by a model. - An illustrator directs forty iterations of a generated image against a specific composition. - A researcher's dataset and findings become an explanatory video narrated by a synthetic voice. **Say it in a sentence** - Short: AI Usage Scale: Level 3 — Directed. - Medium: The knowledge, data and argument are the author's; AI produced the text from them. Level 3 on the AI Usage Scale. - Long: The substance of this work — the knowledge, the data, the argument, the direction — is the author's. A model produced most of the final form from it. The author directed, verified, and takes responsibility for what it says. This is Level 3 on the AI Usage Scale. **Mappings** - Proposed custom `ai-usage`: `3` (not a registered standard) - Experimental `ai-disclosure`: `ai-generated` (not a W3C standard) - IPTC digital source type: `http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia` - schema.org: `aiUsageLevel: 3` - Note: IPTC cannot distinguish Level 3 from Levels 4 and 5. That gap is why this scale exists. --- ### Level 4 — Prompted **AI produced the substance and the form. You asked, you checked, you are accountable.** A model produced both what the work says and how it says it, from a prompt, template, or brief that carried little original, work-specific substance of your own. A human meaningfully reviewed the finished work, checked material claims where necessary, and takes responsibility for publishing it. **This is the level when** - Asking a model for content on a topic and publishing it after review - Generated marketing copy, product descriptions, or social posts, reviewed before going out - Stock imagery generated from a short prompt and checked before use - AI-written code you read and tested before merging **This is not the level when** - Work built on knowledge, data, or direction that is genuinely yours — that is Level 3 - Publishing without a human reading it — that is Level 5 **Examples** - "Write ten posts about diabetes" — read, corrected, published. - Product descriptions generated for a catalogue and reviewed by a person. - A generated header image, checked for errors before it ships. - A summary of a document produced by a model and verified against the source. **Say it in a sentence** - Short: AI Usage Scale: Level 4 — Prompted. - Medium: Produced by AI from a prompt; reviewed by a person who takes responsibility for it. Level 4 on the AI Usage Scale. - Long: A model produced both the substance and the form of this work. A person asked for it, read it, checked it, and takes responsibility for publishing it. This is Level 4 on the AI Usage Scale. **Mappings** - Proposed custom `ai-usage`: `4` (not a registered standard) - Experimental `ai-disclosure`: `ai-generated` (not a W3C standard) - IPTC digital source type: `http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia` - schema.org: `aiUsageLevel: 4` - Note: IPTC cannot distinguish Level 4 from Levels 3 and 5. --- ### Level 5 — Automated **AI produced the substance and the form. No human read it before it was published.** A model produced the work and it was published without human review. This is the honest level for automated pipelines, bulk generation, and agents that publish on their own. It is the right answer for a great deal of legitimate work — and stating it is what makes that work legitimate. > Level 5 is not an accusation. An automated market report, a generated match summary, a machine-translated help page at scale: all of these are honest at Level 5 and dishonest at any other. Declaring 5 is what separates an automated system from a deception. **This is the level when** - Scheduled or agentic publishing with no human in the loop - Bulk generation of pages, listings, or summaries - Live, on-demand generation shown to a user as it is produced - Machine translation published at scale without review **This is not the level when** - Anything a person actually read before it went out — that is Level 4 **Examples** - An hourly market summary generated and posted by a script. - Ten thousand product pages generated overnight. - A chatbot's answers, shown to the user as they are generated. - An agent that writes and publishes to a blog on a schedule. **Say it in a sentence** - Short: AI Usage Scale: Level 5 — Automated. - Medium: Generated by AI and published without human review. Level 5 on the AI Usage Scale. - Long: A model produced this work and it was published without a human reading it first. This is Level 5 on the AI Usage Scale, the honest declaration for automated systems. **Mappings** - Proposed custom `ai-usage`: `5` (not a registered standard) - Experimental `ai-disclosure`: `autonomous` (not a W3C standard) - IPTC digital source type: `http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia` - schema.org: `aiUsageLevel: 5` - Note: IPTC cannot distinguish Level 5 from Levels 3 and 4. --- # The decision procedure Five yes/no questions, asked in order. The first that terminates gives the level. **1. Was any generative AI used in making this work?** Generative AI means a model that produces new text, images, audio, video, or code. Count AI used for research, ideation, or feedback even when its output does not appear in the work. Spellcheck, autofocus, noise gates, search engines, and calculators are not generative AI. - Yes → go to the next question - No → **Level 0** **2. Did AI produce new material that appears in the published work — beyond mechanically processing what you had already made?** Correcting your grammar, faithfully transcribing or translating your words, denoising your photo, or tagging your files is mechanical processing. So is research, ideation, or feedback that leaves no model-produced material in the work. Writing a new paragraph, drawing a new image, or composing a new melody is new material. - Yes → go to the next question - No → **Level 1** **3. Did you make most of the final form yourself — most of the words, pixels, or sound?** Judge the published artifact, not the effort. If most of what a reader sees or hears was typed, drawn, or played by you, answer yes. - Yes → **Level 2** - No → go to the next question **4. Is the substance yours — the knowledge, the data, the argument, the direction?** Substance is the work-specific knowledge, evidence, argument, composition, or constraints that determine what the work says. It is yours when your original inputs materially determine the result — not merely because you wrote or refined the prompt. - Yes → **Level 3** - No → go to the next question **5. Did a human review this before publication and take responsibility for what it says?** Review must be meaningful and appropriate to the stakes: checking the finished work, verifying material claims where necessary, and correcting problems before publication. Taking responsibility means you will defend or correct it if it is wrong. - Yes → **Level 4** - No → **Level 5** --- # Surfaces Optional. A single number describes the work as a whole. When parts of it differ sharply, declare them. The badge always shows one number; the breakdown lives behind the link. **Headline rule.** The headline level describes the part of the work that carries its meaning. Decorative material that a reasonable audience would not consider material to the work does not drive the headline — declare it here instead. Recognised surfaces: text, image, audio, video, code, data. ```json { "aiUsageScale": "1.0", "level": 3, "surfaces": { "text": 3, "image": 5, "audio": 0 }, "note": "Research and argument by the author; images generated." } ``` --- # Edge cases Every one of these came from trying to break the decision tree. **Translation.** A faithful translation inherits the level of the source work and adds a translation note. Translating a Level 1 article with a model publishes it as Level 1, translated by AI. *Why:* The substance and structure are unchanged. Penalising translation would make the world's writing less available, which is the opposite of the point. **Transcription.** Transcribing speech you recorded is mechanical processing. It does not raise the level. *Why:* The words are yours. The model only wrote them down. **Decorative assets.** A generated header image on a human-written article does not make the article Level 4. Declare the image under Surfaces. *Why:* The headline level describes what carries the meaning. Otherwise one stock image would swallow the whole declaration, and people would simply stop declaring. **Derivative works.** Declare the level of your own contribution, and cite the level of the source if it declared one. **Live generation.** Content generated on demand and shown to a user without review is Level 5, even if the system prompt was written carefully by a person. **Non-generative machine learning.** Autofocus, denoise, upscaling, colour matching, and classification that invent nothing are not generative AI. They do not move you off Level 0. --- # Interoperability **W3C AI Content Disclosure Community Group** (https://www.w3.org/community/ai-content-disclosure/) The W3C AI Content Disclosure Community Group is discussing candidate syntax and a four-part model, but has not published a W3C standard. This scale currently emits experimental ai-disclosure metadata aligned with that model; consumers must not treat it as standardised W3C markup. **IPTC Digital Source Type NewsCodes** (https://cv.iptc.org/newscodes/digitalsourcetype/) Every level maps to an IPTC term, so a declaration can travel into C2PA manifests, XMP, and the metadata Meta, Pinterest, and Google already read. The mapping is lossy in one direction only: IPTC cannot tell Levels 3, 4, and 5 apart. **C2PA / Content Credentials** (https://c2pa.org/) Complementary, not competing. C2PA provides tamper-evident, signed provenance about an asset and its processing history. Its core specification does not attribute content to individuals or organisations, and provenance alone cannot determine whose thinking shaped a work. This scale is a declaration about contribution and review, designed to work for web text as well as media files. ## Level mapping | Level | Name | Experimental ai-disclosure | IPTC digital source type | |---|---|---|---| | 0 | Human | `none` | `digitalCreation` | | 1 | Assisted | `ai-assisted` | `algorithmicallyEnhanced` | | 2 | Co-created | `ai-assisted` | `compositeWithTrainedAlgorithmicMedia` | | 3 | Directed | `ai-generated` | `trainedAlgorithmicMedia` | | 4 | Prompted | `ai-generated` | `trainedAlgorithmicMedia` | | 5 | Automated | `autonomous` | `trainedAlgorithmicMedia` | The mapping is lossy in exactly one direction, and the loss is the point: IPTC cannot tell Levels 3, 4 and 5 apart. It has no term for *whose substance this is*, and none for *whether a person read it*. Those are the two questions readers actually care about. --- # Prior art **AI Assessment Scale (AIAS)** — https://aiassessmentscale.com/ · Perkins, Furze, Roe, MacVaugh Five non-hierarchical levels, adopted by hundreds of institutions in 30+ languages. AIAS describes what a student is permitted to do. This scale describes what an author did. The debt is direct: the principle that no level ranks above another is theirs, and it is the most important rule here. **Creative Commons** — https://creativecommons.org/ The three-layer model — a mark anyone can read, a page that explains it, a specification that pins it down — is theirs. **Human Provenance in Film** — https://humanprovenance.film/ Three tiers, free, CC BY 4.0. Proof that a graded standard can launch in a hostile industry. --- # How to use it 1. Answer the five questions. Take the number. 2. Put the mark on the work, linked to `https://usagescale.org/`, or just write the sentence. 3. If different parts of the work differ sharply, declare the surfaces too. Machine-readable, in the ``: ```html ``` These fields describe provenance and review. They do not grant or deny permission to use the work for training; licences, terms and access controls do that. There is no fee, no account, no registry, and no certification. The declaration is public and linkable, which makes it falsifiable in public. Reputation is the enforcement, and it is the only one that scales — detection is not a backstop, and never was: AI detectors flagged 61% of genuine essays by non-native English speakers as machine-written. --- # This document declares its own level **Level 3 — Directed.** The diagnosis, the argument and every design decision in the scale are the author's. The research and the prose were produced with a large language model, then read, corrected and signed line by line. That is exactly the case this standard defends. Hiding it would have been absurd.