Trained On AI training-data disclosure
← Register

Black Forest Labs — FLUX 3

What Black Forest Labs has published for FLUX 3 under Article 53, and when we last looked. Every row links to the page the value was read from.

Unverified checked today
Provider Black Forest Labs
Black Forest Labs publishes this page to provide regulatory information about how our models are developed, trained, and governed.
source · checked 2026-09-01
GPAI provider Yes
our General-Purpose AI models
source · checked 2026-09-01
Code of Practice Not stated
source · checked 2026-09-01
AI Act statement Open document bfl.ai
Black Forest Labs publishes this page to provide regulatory information about how our models are developed, trained, and governed.
source · checked 2026-08-16
Incident contact Not stated
source · checked 2026-09-01
California AB 2013 disclosure Published
Training data disclosure for the family of FLUX multimodal models, published pursuant to California Civil Code § 3111 (AB 2013).
source · checked 2026-08-16 · human-verified
AB 2013 disclosure URL Open document bfl.ai
source · checked 2026-08-16 · human-verified

What we could not read

One source for this model could not be read on the last check. Any blank above may be a limit of our monitoring rather than a fact about the provider.

We do not disguise this checker as a browser to get past a block. This register publishes claims about the companies whose access controls it would be circumventing, so a refused source goes to manual review instead — which is slower, and occasionally leaves a cell empty that a person could fill. If you can see a document for this model that we cannot, send it to hello@trainedon.org and it will be recorded with your source.

If you are building on this model

Using a general-purpose model inside your own product generally makes you a downstream provider with obligations of your own — and those do not transfer from the model's provider. This is what the register could find of the material you would normally gather, and what to request where it could not. 0 of 5 available publicly for this model.

  • Training-data summary Not found here

    Article 53(1)(d) — the public summary of training content.

    Ask the provider for the public summary of training content, or its URL.

  • Copyright policy Not found here

    Article 53(1)(c) — the policy for complying with EU copyright law.

    Ask for the copyright policy, including how text-and-data-mining reservations are honoured.

  • Documentation for downstream providers Not found here

    Article 53(1)(b) — the information you need to meet your own obligations.

    Request the Annex XII information package. This is the one you most likely need in writing.

  • A stated systemic-risk position Not found here

    Article 55 — changes what is expected of the model and what you should ask about.

    Ask whether the provider has notified the Commission under Article 52. No provider on this register states a position publicly.

  • A contact for incidents Not found here

    Somewhere to report a serious incident you observe in production.

    Establish a named contact before deployment, not after an incident.

Availability, not compliance. A blank means this register did not find the document at the sources it monitors — which may be a gap in our monitoring rather than a fact about the provider. Whether any obligation applies, and to whom, is a legal question this register does not answer.

Source documents archived

We keep a dated copy of every source page, captured whenever its content changes. 1 document is held for this model, watched since .

Vendor pages are edited in place and are not reliably captured elsewhere, so a claim made today cannot otherwise be checked against the page as it stood. Ask for a copy at hello@trainedon.org.

Cite this entry

Every claim here is dated on purpose. Please keep the date attached.

Trained On, "Black Forest Labs — FLUX 3", GPAI transparency register, checked 2026-09-01. https://trainedon.org/flux-3/

Think this is wrong or out of date? Send us the URL — corrections are applied within a week and the change is recorded in the change log.