The training layer: a meta-aggregator / Have I Been Trained?
Have I Been Trained?
What it is
Spawning was founded by the artists Mat Dryhurst and Holly Herndon with the stated goal of giving creators more control over how their work is used in AI training. A search takes text or an uploaded image and returns similar images from the dataset through OpenAI's CLIP model, each with a similarity score, its caption in the dataset and the URL it was scraped from.
From its /non article, sentences 4, 6, 7
Beside the search Spawning keeps a Do Not Train registry, a global opt-out list said in 2026 to hold over 80 million entries, which its reviewers call a consent signal without an enforcement mechanism.
From its /non article, sentences 17, 18
What it cannot see
The search covers LAION-5B alone: none of the text training data of language models and none of the proprietary image data of Midjourney or OpenAI. An exact image found proves it was in the training data; a similar one proves much less, and not finding a work shows only that it is not in LAION-5B.
From its /non article, sentences 11, 8, 12
The archive names what such a search cannot reach training inscription: a work can persist in a model as a constraint on what the model does without remaining an inspectable copy.
From its /non article, sentence 14
This article is derived from the entity's /non article (EA-NEGONT-02, the knowledge object composed under v0.10, #1726): every fact in it is stated there, and each paragraph names the sentences it rests on.
The census record
| Relation to this census | infrastructure |
|---|---|
| Record type | service |
| The question it answers | consent and discovery: can a work be found in training data and withheld from training through a Do-Not-Train registry? (present state read-needed: the site served a maintenance page) |
| Kind | not recorded in the saved sources: on 2026-11-01 haveibeentrained.com served a static Spawning maintenance page that names a Do-Not-Train registry |
| Operator | Spawning (the wordmark on the page served at haveibeentrained.com) |
| Evidence | opt-out registry |
| Unit of record | not recorded |
| What it records | not recorded; the served page names a Do-Not-Train registry and addresses 'AI trainers looking to respect' it |
| In the layer | not recorded; the page names a Do-Not-Train registry that AI trainers may respect, which bears on what enters training, and says nothing of whose works are in it |
| Under whose name | not recorded |
| Found and read back | not recorded; on 2026-11-01 the home page and the FAQ URL both served the same maintenance page, and no search was available to read |
| What the apparatus makes from it | not recorded |
| Method | not recorded |
| Coverage | geography: not recorded; period: not recorded |
| Count | value: not stated in the saved sources; as_of: 2026-11-01 |
| Holdings | a Do-Not-Train registry, named on the maintenance page |
| Stated limits | the site states it is 'Under Maintenance' |
| Data access | none on 2026-11-01: the site served a maintenance page; contact by email for AI trainers (the address is obscured in the saved text) |
| Licence | not recorded |
| Lifecycle | state: paused; detail: on 2026-11-01 haveibeentrained.com, its /faq URL and spawning.ai each served a static 'Under Maintenance' page with the Spawning wordmark (the saved HTML is plain markup and states that no JavaScript is used); the sources were refetched at 05:07 UTC on 2026-11-01 and still served the same page; when the maintenance began, and what the site served before, is not recorded |
| Version and time | maintenance_observed: 2026-11-01; refetched: 2026-11-01T05:07Z (manifest); read: 2026-11-01 |
What its record can settle
- that on 2026-11-01 haveibeentrained.com and its FAQ URL served a Spawning maintenance page
- that Spawning names a Do-Not-Train registry and invites AI trainers who would respect it to make contact
What it cannot settle
- whether a named work or image is in any training set (no search was served)
- what the Do-Not-Train registry holds, how a work is entered or how trainers apply it
- whether any trainer respects the registry
Kept apart
- NOT readable as a search tool on the day read: it served a maintenance page
- NOT a statement that any trainer honours the Do-Not-Train registry: the page invites trainers 'looking to respect' it to make contact
- NOT a script-rendered page left unread: the saved HTML is a static maintenance notice
- NOT a literary corpus: discovery and consent infrastructure, read on a day it served a maintenance page
- NOT the Do-Not-Train registry itself: the registry is a separate object named on the page
Captures: the composition layer's account
- «Have I Been Trained» on Google AI Overview (expanded), 2026-11-01the composition layer's account of the entity at its name, seated in the Capture Registry (AXN:0BB4)
The /non row
Evidence
- maintenance (the resource's own statement):
Spawning Under Maintenance Hacking the mainframe...
https://haveibeentrained.com/ · read 2026-11-01 · http 200 · sha256 40dc9cb4a48518b8… - registry (the resource's own statement):
AI trainers looking to respect the Do-Not-Train registry, please reach out to us at
https://haveibeentrained.com/ · read 2026-11-01 · http 200 · sha256 40dc9cb4a48518b8… - faq_same_page (the resource's own statement):
Spawning Under Maintenance Hacking the mainframe... AI trainers looking to respect the Do-Not-Train registry
https://haveibeentrained.com/faq · read 2026-11-01 · http 200 · sha256 6ca4efeb6186770e…
Unchecked
the present state of the service: read-needed; the refetch of 2026-11-01 05:07 UTC served the maintenance page again; the site's search and FAQ: on 2026-11-01 both URLs served the maintenance page, so the search, its index, its opt-out procedure and its FAQ text are unread; spawning.ai (spawning-hibt--spawning): 71 characters of text, below the census's reading threshold and unquoted; its saved HTML is titled 'Source.Plus' and serves the same maintenance notice; the saved HTML of haveibeentrained.com carries a meta description, 'Data Governance for Generative AI', outside the saved text and so not verbatim-checked here; the name 'Have I Been Trained?' does not appear in the saved text; it is taken from the domain
Cite: Lee Sharks, ed. "Have I Been Trained?." The training layer: a meta-aggregator, traininglayerliterature.org, compiled 2026-11-01. https://www.traininglayerliterature.org/aggregators/spawning-hibt