BlendedBoris Blog: Simple Blockchain Ownership Registration

OpenAI Copyright Lawsuit: What Midjourney's Fair Use Brief and the Photographers and Illustrators Lining Up Against It Mean for Blender Artists (2026) | Blended BORIS | Digital Art IP

By the 30 October 2026 deadline, more than 50 groups and people had asked a New York federal court for permission to weigh in on the biggest AI copyright case in the United States. Among them were two AI image and video companies, Midjourney and Runway, which told the judge that training AI on the internet's images is fair use. On the other side were photographers, stock image licensors, medical illustrators and a consent group backed by Cate Blanchett. For Blender artists whose renders, models and textures sit on public portfolios, this is the fight that may decide whether that work can be used for AI training without asking.

The case is In re OpenAI, Inc. Copyright Infringement Litigation, a combined proceeding before Judge Sidney H. Stein in the Southern District of New York. It brings together lawsuits by The New York Times, other news publishers and book authors against OpenAI and Microsoft. On 4 September 2026, both sides asked the judge to decide the fair use question without a trial. The publishers and authors say copying their work to train ChatGPT was infringement. OpenAI and Microsoft say it was fair use.

The case is about articles and books, not 3D art. But it is the most important pending US test of whether AI training is fair use, and the arguments now on file speak directly to images, artists and licensing. This article explains what was filed, who said what, what the judge has to decide and what Blender artists can do while they wait.

TL;DR: the OpenAI copyright lawsuit and Blender artists

Techdirt counted 53 proposed amicus briefs filed by 30 October 2026: 22 for OpenAI and Microsoft, 26 for the publishers and authors, and 4 for neither side. Midjourney and Runway argued that training on images is fair use. Photographers, illustrators and stock licensors argued the opposite, and the Human Consent Foundation said the lack of an opt-out tag is not consent. Replies are reported as due on 20 November. No hearing date is set and nothing has been decided. Whatever the outcome, evidence matters: dated source files, records of lost work and licences, and opt-outs you can prove you set.

What is happening in the OpenAI copyright case?

Summary judgment is a request to decide a case, or part of it, on the written record because the key facts are not in real dispute. Here, OpenAI, Microsoft and the news plaintiffs all filed such motions on 4 September 2026. Oppositions followed on 23 October. Under the schedule reported by IPWatchdog, reply briefs are due on 20 November 2026. No hearing date has been announced, and there is no deadline for a ruling.

Fair use is the US rule that allows some unlicensed use of copyrighted work. Courts weigh four factors: the purpose and character of the use, the nature of the work, how much was taken and the effect on the market for the original. The first and fourth factors usually decide AI cases. Is training a new purpose that transforms the work, or does it build a product that competes with it?

The US government has already taken a side. On 1 September 2026, the Justice Department filed a statement of interest arguing that training large language models on copyrighted text is fair use. A statement of interest is not binding on the judge. On 30 October, OpenAI published a post called "Fair use benefits everyone" that quoted 20 filings supporting its position, as IPWatchdog reported.

Who filed amicus briefs, and on which side?

An amicus brief is a filing by someone who is not a party but wants the court to consider their view. In this case each amicus had to ask the court for leave first, so all of these briefs are proposed until the judge accepts them. Mike Masnick of Techdirt sorted the briefs on the docket up to 30 October: 22 support the defendants, 26 support the plaintiffs, 4 support neither side and 1 had not been located yet. Techdirt filed one of the pro-fair-use briefs itself, through its Copia Institute.

On the fair use side are tech trade groups, startups, libraries, the Electronic Frontier Foundation, several law professors, the venture firm Khosla Ventures and two independent creators who use AI. The Authors Alliance, a group of authors that supports fair use, filed on one narrow point: that "market dilution", the idea that AI output floods the market and lowers the value of human work, is not a harm copyright law recognises.

On the other side are publishers, the music industry, the Copyright Alliance, the Writers Guilds, SAG-AFTRA and the American Federation of Musicians, the talent agency WME, Ken Burns and a group of copyright law professors. Several briefs come from visual creators, which we cover below.

What did Midjourney and Runway tell the court?

Midjourney, maker of the image generator of the same name, and Runway, which builds video generation models, filed a joint proposed amicus brief late on 30 October in support of OpenAI and Microsoft. They told the court they are not parties and have no stake in the outcome. They argued that generative AI is a catalyst for human creativity and that training models on the world's writing and images is a classic fair use.

The brief explains that image and video models are trained on billions of images and videos paired with text descriptions, and says they learn general visual ideas, such as what a car, a flower or a cat looks like, rather than how particular artists depicted them. It says these tools were trained on writings, photographs and artworks available on the public internet, which it called "our collective birthright to learn from and build upon". Most of the brief is a list of examples of artists, designers and disabled creators who use AI tools.

For artists, two points stand out. First, this is the first time in this case that image and video AI makers have spelled out how they see training on visual work, and their answer is that it needs no permission. Second, Midjourney is itself a defendant in other suits, including the artists' class action Andersen v. Stability AI and a suit brought by Disney and Universal. We looked at one of the reports behind those concerns in our article on Midjourney and AI character copies.

Whatever the court decides, your date is yours.

BlendedBoris records a blockchain timestamp of your Blender work from inside Blender, so you hold dated proof of when you made each model, sculpt and render.

👉 Protect your Blender work with BlendedBoris

What did photographers, illustrators and consent groups argue?

A coalition of writers' and photographers' groups, including the American Society of Media Photographers and the National Press Photographers Association, backed the authors. Their brief argues that AI tools are flooding the market with substitutes for human work. It cites a 2026 survey by the UK Association of Photographers that reported a 65% drop in commissioned images being licensed by its members compared with the year before, and a 46% drop in images shown on photographers' own websites, which the survey linked to fears of scraping.

A second coalition, led by the AI Coalition for Data Integrity, includes the Digital Media Licensing Association, which represents stock image libraries, the Association of Medical Illustrators and a microstock company. Its brief answers the government's national security argument: security agencies want to know where training data comes from, and only licensed data with a clear record of origin gives that.

The third brief to watch comes from the Human Consent Foundation, joined by Cate Blanchett, Steven Soderbergh, the agency CAA and others. The foundation maintains the Human Consent Standard, a machine-readable way for creators to state how AI may use their work and likeness, including visual art. Its brief argues that a working market for licensing training data already exists. It also makes a point every artist should note: an opt-out system only works one way, because, in the brief's words, "technical silence is not legal consent". In other words, a missing NoAI tag should not be read as permission.

That fits what we reported on the new shared robots.txt signal for AI training and on the NoAI tag ruling in the Roblox case. Opt-out tags help, but they are a request, not a licence, and how much legal weight they carry is still being argued in court.

Why does a case about news articles matter for 3D artists?

Fair use rulings travel. Judge Stein's answer will be cited in every AI training case in the US, including the artists' cases against image generators. The two big 2025 rulings from California show how much depends on the details. In Bartz v. Anthropic, Judge Alsup found that training on books was fair use but that keeping pirated copies was not, and the case later settled. In Kadrey v. Meta, Judge Chhabria ruled for Meta because the authors had not proved market harm, while warning that market dilution could win in another case. In September 2026, the Third Circuit went the other way on a non-generative tool, as we explained in our report on the Thomson Reuters v. ROSS appeal.

Three questions in the briefs matter most for visual artists. Does training on public images count as a new, transformative purpose? Does a flood of AI images and models count as market harm? And does an existing market for licensing training data, from stock libraries to paid dataset deals, show that AI companies should pay? Our report on the open-source Hi3D image-to-3D generator showed how 3D models from public platforms already end up in training sets, and Sketchfab's new terms of use show how platforms describe AI use of uploads.

Filer	Side	Main argument	What it means for Blender artists
US Justice Department (1 Sep)	Fair use	Training language models is fair use	Not binding, but the judge may weigh it
Midjourney and Runway	Fair use	Training on public images is quintessential fair use	Image and video AI makers say renders need no permission
Authors Alliance, CCIA, law professors	Fair use	Market dilution is competition, not harm	Lost sales to AI output may not count
Photographers (ASMP, NPPA) and writers	Not fair use	AI floods the market with substitutes	Records of lost commissions become evidence
Stock and medical illustrators (DMLA, AMI)	Not fair use	Licensed data gives provenance	Licensing your work is a real market
Human Consent Foundation, Blanchett, CAA	Not fair use (no explicit ask)	Licensing works; silence is not consent	A missing NoAI tag should not mean yes
Wikimedia Foundation	Neither	Judge each use separately	Training and later uses may be treated differently

This is US law. In the European Union, AI developers can mine content unless the rights holder has reserved its rights in a machine-readable way, so opt-outs carry direct legal weight there. Our guide to whether 3D models automatically have copyright explains why your models are protected from the moment you make them, wherever you live, and our list of common copyright problems for 3D artists covers what usually goes wrong.

What should Blender artists do now?

Keep your source files. Save .blend files, sculpts, texture projects and early versions. They show that a model or render started with you.
Date your work. Timestamp key versions as you go, so you can show when each piece existed before it was scraped or copied.
Register key works in the US. If you sell in the US, registering early keeps statutory damages and legal fees available. See how to register copyright in 2026.
Set your opt-outs and record them. Use NoAI settings where platforms offer them and AI training rules in your own site's robots.txt. Take dated screenshots. Our steps to opt out of AI training datasets walk through the options.
Read platform and licence terms. Check what each portfolio site and marketplace says about AI training before you upload, and whether your licences say anything about it.
Record lost work and licence offers. Note commissions that dried up, clients who switched to AI and any offer to license your work for training. Market harm and licensing markets are argued with evidence like this.
Watch for outputs that copy your art. Save examples of AI images or models that reproduce your characters or designs, with dates and the prompts used.
Keep proof in one place. Store files, timestamps, screenshots and records together so you can use them in a takedown, a claim or a licensing talk.

The sister blog explains how fair use works for digital art, what rights artists have over AI training and which court decisions artists should know. If you want to compare how other courts are treating AI and authorship, see the Frankfurt court's 'it was AI' ruling and the Gruffalo illustrator's possible case over ChatGPT copies.

Where does this fit in the bigger picture?

AI is changing who can copy what, and how fast. This week our sister sites covered the same shift from the engineering and security side:

AI agents as attack tools. CrowdStrike and Google's threat team reported in late October 2026 that espionage groups used an AI agent hacking framework called ARTEX to break into organisations far faster than before: what the ARTEX AI cyber espionage reports mean for CAD data on cadchain.com.
Ransomware on design storage. Synology warned on 29 October that attackers are encrypting outdated network storage devices and demanding payment: what the Synology NAS ransomware wave means for SOLIDWORKS files on solidboris.com.
Broken trust in certificates. Eight eMudhra intermediate certificates were revoked after a key compromise, and Microsoft shipped updates to remove trust: what the eMudhra certificate compromise means for Inventor offices on borisinventor.com.

The common thread is proof. Whether the question is who made a model, who copied a file or which certificate can be trusted, the person with clear, dated records is in the strongest position. For artists who license their work, our report on Getty Images' Chapter 11 filing looked at what changes for 3D artists who sell stock.

Conclusion

The OpenAI copyright lawsuit is not about Blender, but its outcome will shape how AI companies treat Blender artists' work. Midjourney and Runway have now said plainly that they see training on public images as fair use. Photographers, illustrators, stock licensors and consent groups have answered that a licensing market exists and that silence is not consent. Judge Stein has not ruled, replies are reported as due on 20 November and appeals are likely whatever he decides. In the meantime, the practical answer for artists is the same: keep dated originals, set and record your opt-outs, register key works and keep evidence of what AI costs you.

Your renders, your date, your proof.

BlendedBoris adds a blockchain timestamp to your Blender work from inside Blender, giving you dated proof of authorship to stand next to your copyright registrations, licences and opt-outs.

👉 Get started with BlendedBoris

FAQ: the OpenAI copyright lawsuit and Blender artists

What is the OpenAI copyright lawsuit about?

It combines suits by The New York Times, other news publishers and book authors against OpenAI and Microsoft in a New York federal court. They say training ChatGPT on their work without a licence infringed copyright. OpenAI and Microsoft say it was fair use.

Has the judge ruled on fair use?

No. Both sides moved for summary judgment on 4 September 2026, oppositions were filed on 23 October and replies are reported as due on 20 November. No hearing date or ruling date has been set.

Why did Midjourney and Runway file a brief?

They filed a proposed amicus brief on 30 October 2026 supporting OpenAI and Microsoft. They argued that training AI on public writing and images is fair use and that AI tools help people create.

Did any artists file against OpenAI?

Yes. Photographers' groups, stock image licensors, medical illustrators, writers' unions and a consent coalition with Cate Blanchett filed briefs supporting the publishers and authors.

What is market dilution?

It is the argument that AI output floods the market and lowers the value of human work, even if no single output copies a specific work. Some briefs say it is real harm. Others say it is just competition.

Does a NoAI tag protect my work in the US?

It records your refusal, but its legal weight is still being argued. The Human Consent Foundation's brief argues that a missing tag is not consent. In the EU, machine-readable reservations carry more direct legal weight.

Does this case decide anything for 3D models?

Not directly. It is about text. But its fair use reasoning will be cited in the image and 3D cases that follow.

How can BlendedBoris help?

BlendedBoris records a blockchain timestamp of your Blender work from inside Blender, giving you dated proof of when you made each version of a model or render.

Violetta Bonenkamp, CADChain co-founder

2026-11-03 16:25 News