DataPacts

Stability AI: Data Buyer Profile, Program Terms, Payouts and Reputation

Checked

At a glance

Stability AI, the image and audio model developer registered in England as STABILITY AI LTD, has agreements for AI music tools with Universal Music Group (2025-10-30) and Warner Music Group (2025-11-19) and states that Stable Audio 3.0 is trained on fully licensed data, while no program for companies that want to license data to it appears in the sources checked as of 2026-11-09. Neither label disclosed a value.

Independent reference. Every buyer is covered on the same terms, every figure carries its source and check date, and no buyer pays to be listed or placed higher.

Request an independent introduction

Tell us what data your company holds. We reply with the buyers that fit and what their terms leave out. Your details reach a buyer only after you agree to a named introduction.

IndependentYour details stay with us until you agreeNot selling is a valid outcome

The same company is a defendant in training-data lawsuits in the UK and the US, with no final ruling on whether training on the claimants' works was lawful, as of 2026-11-16. Every figure here is a dated statement by Stability, its counterparty, a register, a court or a regulator page, collected by the same method as every buyer profile in this independent reference on AI training data licensing.

Each number carries its source and check date, and how DataPacts sources and dates every figure is set out in the methodology.

What Stability AI Acquires on Record: Music Catalog Agreements and Licensed Audio

Two acquisition routes appear in the record for Stability AI, both in music: agreements with two major labels and a model family Stability says is trained on fully licensed data.

  • Label agreements for music tools: Universal Music Group and Warner Music Group.
  • Licensed data behind Stable Audio 3.0, as Stability states it.
  • Image training data: only court allegations and findings, no Stability statement.
  • Company records: no program on record.

Not on record are a named image licensor, a purchased dataset, a seller program or a Stability crawler.

The difference matters to a holder weighing a licensee whose image models are in litigation. A grant for music tools or a licensed-audio model is a different bargain from a grant for image training, and the line between licensed versus scraped training data decides what a licensor can ask the licensee to confirm.

Stability AI Company Identity

The contracting party behind Stability AI is STABILITY AI LTD, a company registered at Companies House under number 12295325, active and incorporated on 2019-11-04 (checked 2026-11-09).

Attribute Value Source
Legal name Stability AI Ltd; brand Stability AI Companies House
Register entry and status STABILITY AI LTD, 12295325, active Companies House
Incorporated 2019-11-04 Companies House
Website stability.ai stability.ai
Country and type United Kingdom; frontier lab, image and audio model developer Companies House
US defendant entities Stability AI, Inc. and Stability AI US Services Corporation (named in the US complaint) N.D. Cal. complaint, 3:25-cv-06891-TLT
Founded, founders, chief executive, headcount, registered office not recorded n/a

Companies House identifies the UK entity, and the US action also names Stability AI, Inc. and Stability AI US Services Corporation, so a contract should name the exact signing entity.

The date is an incorporation date, not a founding date. For the licensing frame that applies to a UK licensee, see the page on rules for licensing data for AI training in the United Kingdom.

Stability AI Program Terms and Payouts, Dated

Stability AI publishes no program, application page or terms for licensing company data; the table lists each term a data holder would check and what the record holds, as of 2026-11-09.

Term What the record shows Checked Source
Company data program or application page none on record 2026-11-09 stability.ai
License scope, term, exclusivity and deletion for company data none published 2026-11-09 stability.ai
Payout to data holders none stated 2026-11-09 stability.ai
Referral fee none recorded 2026-11-09 stability.ai
Agreements on record UMG alliance for AI music creation tools (2025-10-30, value not disclosed) and WMG agreement for AI tools (2025-11-19, value not disclosed; WMG calls its AI deals "variable economics") 2026-11-09 stability.ai; sec.gov WMG Ex. 99.1
Licensed-data statement Stable Audio 3.0 "trained on fully licensed data" (Stability's claim, 2026-05-20) 2026-11-09 stability.ai
Opt-out mechanism and crawler none recorded 2026-11-17 n/a
GPAI Code of Practice not among the 22 listed signatories (Commission page updated 2026-10-07) 2026-11-09 European Commission
EU training content summary none recorded 2026-11-09 n/a

Every company-data row reads "none on record". That leaves no scope, term, exclusivity, deletion or payment trigger to compare, and the rows are read against the program terms compared across buyers that do publish them. The wording does not say Stability never acquires data.

The label agreements are the only terms on record. Neither party disclosed a value, and WMG describes its AI deals as based on "variable economics".

Stability AI's Label Agreements: Tools, Investment and What Is Not Disclosed

Two agreements with record labels name Stability AI as the counterparty, Universal Music Group on 2025-10-30 and Warner Music Group on 2025-11-19, and neither party disclosed a value.

The table lists the agreements and the equity round, newest first, as of 2026-11-09, with values as stated by the parties.

Date Counterparty What is stated Value Source
2026-08-25 Sony Music Group, Universal Music Group, Warner Music Group, Electronic Arts (among investors) Series B equity round; total funding USD 232M (Stability's claim); not a license not a license fee stability.ai
2025-11-19 Warner Music Group AI tools for music creation; WMG 8-K exhibit of 2026-03-03: deals "based on variable economics" not disclosed stability.ai; sec.gov WMG Ex. 99.1
2025-10-30 Universal Music Group strategic alliance to co-develop professional AI music creation tools not disclosed stability.ai

"Tools" is the parties' word. Neither announcement is recorded as a training-data license, and that does not mean training is excluded.

The WMG agreement covers AI tools for music creation. In its 8-K exhibit of 2026-03-03, WMG lists agreements with Suno, Udio, Stability AI and Klay and says all its deals are based on variable economics. The wider market for catalogs is covered for musicians, labels and publishers licensing music for AI training.

Stability states a Series B on 2026-08-25 with Sony Music Group, Universal Music Group, Warner Music Group and Electronic Arts among the investors and total funding of USD 232M. That is an equity investment, not a license fee, and no round size is stated here.

Stability AI Training Data Statements, Opt-out and EU Status

Stability AI's only training-data statement on record concerns audio: it states that Stable Audio 3.0, released on 2026-05-20, is trained on fully licensed data.

That is Stability's claim, and no licensor has confirmed it in the record.

Stability makes no statement on the sources of its image training data in the record. The UK judgment cites LAION-5B at 5.85 billion URL-text pairs (paragraph 47), and the US complaints allege training on Getty images and LAION-based data; these are findings or allegations, not Stability statements.

No opt-out mechanism, registry or form stated by Stability AI and no Stability crawler are recorded as of 2026-11-17, so this page names none.

Stability AI is not among the 22 signatories listed for the voluntary GPAI Code of Practice (Commission page updated 2026-10-07, checked 2026-11-09), and no training content summary by Stability is recorded. The GPAI obligations of Regulation (EU) 2024/1689 apply from 2025-08-02, and models placed on the market earlier have until 2027-08-02.

General exclusion routes and their limits are set out on the page about opting out of AI training.

Stability AI as a Defendant: The Training Data Record as of 2026-11-16

Stability AI is the defendant in three training-data cases on record and a fourth filed in 2026, and none has produced a final ruling that training on the claimants' works was lawful or unlawful, as of 2026-11-16.

The table compares the four cases as filed, with court, docket and status as of 2026-11-16.

Case (as filed) Court and docket What is alleged or decided Status Source
UK claim by Getty Images companies High Court IL-2023-000007, [2025] EWHC 2863 (Ch); Court of Appeal CA-2026-000007 training claim abandoned (no evidence of UK training); model not an infringing copy; trade mark infringement in part; permission to appeal on secondary infringement 2025-12-16 appeal heard 2026-11-11 and 2026-11-12, judgment reserved 2026-11-12 caselaw.nationalarchives.gov.uk; casetracker.justice.gov.uk
US action by Getty Images (US), Inc. N.D. Cal. 3:25-cv-06891-TLT (earlier D. Del. 1:23-cv-00135) alleged copying of Getty images to train Stable Diffusion; watermark outputs stayed until 2027-01-13 (Dkt. 81, 2026-10-30) courtlistener.com; archive.org RECAP
Artists' putative class action (Andersen) N.D. Cal. 3:23-cv-00201-WHO alleged infringement through LAION-based training; copyright claims survived 2024-08-12 (Dkt. 223) discovery as of 2026-10-19 (Dkt. 778) archive.org RECAP
EVOX Productions LLC v. Stability AI, Inc. C.D. Cal. 2:26-cv-07201-SVW-DMK copyright claims by a vehicle image library (not summarized) pending, filed 2026-07-02 archive.org RECAP

The High Court judgment of 2025-11-04 held that Stable Diffusion is not an infringing copy and found trade mark infringement in part, and the Court of Appeal heard the appeal on 2026-11-11 and 2026-11-12 and reserved judgment on 2026-11-12 (CA-2026-000007). The full record sits on Getty Images v. Stability AI in the UK.

Getty abandoned its training claim because there was no evidence that training took place in the UK, so no court has decided whether training Stable Diffusion on Getty images infringed.

The appeal concerns secondary infringement only: permission was granted to Getty on 2025-12-16 and refused to Stability on the trade mark findings, and on 2025-12-17 costs went to Stability as overall winner.

The US action, 3:25-cv-06891-TLT, was filed on 2025-08-14 after Getty dismissed its Delaware action without prejudice, and it alleges copying of Getty images to train Stable Diffusion. It is stayed until 2027-01-13 (Dkt. 81), and the page on Getty Images v. Stability AI in the US tracks it.

A trial date of 2028-01-18 was set before the stay, and no settlement is recorded. The US complaint pleads the training-based copying that the UK case never reached, so the two cases are not the same dispute and their outcomes do not transfer.

The artists' action, 3:23-cv-00201-WHO, was filed on 2023-01-13 and alleges infringement through LAION-based training. Its copyright claims survived on 2024-08-12 (Dkt. 223), and it was in discovery as of 2026-10-19 (Dkt. 778); the page on Andersen v. Stability AI holds the detail.

The DMCA claims there were dismissed with prejudice.

A fourth suit, filed on 2026-07-02 in the Central District of California by a vehicle image library, is pending, and only its filing data are recorded here.

No breach notice, regulator action or licensor dispute is recorded in the sources checked, and the dated log of data buyer incidents holds the comparison across buyers.

Benefits, Risks, Downsides and Alternatives for a Data Holder Considering Stability AI

For a data holder, Stability AI offers an identifiable UK licensee with two label agreements and a licensed-audio claim, but no published route or terms for company data and an open litigation record, so the practical alternatives are buyers that publish programs, and the option of not licensing at all.

Benefits (as the record supports them):

  • An identifiable UK entity with an active register entry.
  • A documented history of agreements with two major labels.
  • A public claim of fully licensed training data for Stable Audio 3.0 that a licensor can hold it to.

Risks:

  • No published license terms, so scope, exclusivity, deletion and payment would be negotiated from zero.
  • No value disclosed in either label agreement, so there is no price reference.
  • Open litigation over past image training, with a reserved appeal judgment and a stayed US action.
  • Not a listed GPAI Code signatory, and no training content summary on record.

Downsides:

  • No application route, payout or referral fee for company data.
  • Both agreements on record concern record labels, not company records.
  • No opt-out statement on record for holders who prefer exclusion.

Alternatives (alphabetical; each one attribute as the party's claim, checked 2026-11-09):

  • Bria AI data partners, for image libraries: Bria states "Compensation ties to influence, not catalog size", with no rate published.
  • Defined.ai partnerships: data partners earn licensing fees from marketplace sales, with terms approved by the partner and no revenue share percentage stated.
  • Protege data providers: Protege states it "provides revenue share payouts to data partners with each use", with the percentage not public.
  • Troveo content owners, for video: royalty each time content is used, with no rate published.
  • Wirestock, for individual creators: "You earn 50% commission each time someone licenses one of your assets".
  • Not licensing at all, which keeps the data and every later option open.

These are each party's claims, not payments made. The wider set of alternatives to selling data includes keeping the data.

Stability AI Among Frontier Labs That Acquire Data

Stability AI is one of the frontier labs profiled as data buyers, and one whose agreements on record concern music rather than text or news.

It sits in the comparison of frontier AI labs as data buyers, and the hub on who buys AI training data places it beside every other buyer. When a lab has no company program, choosing a buyer for your data covers the next step.

More About Stability AI as a Data Buyer

Where Stability AI's Agreements and Cases Are Tracked

Each record above has its own tracker page with full terms, sources and status.

Frequently Asked Questions About Stability AI and Training Data

Does Stability AI pay for training data?

No payment amount is on record: Stability AI's agreements with UMG and WMG disclose no value, WMG calls its AI deals variable economics, and Stability states Stable Audio 3.0 uses fully licensed data.

Is Stable Audio trained on licensed data?

Stability AI states that Stable Audio 3.0, released on 2026-05-20, is trained on fully licensed data. That is Stability's claim, and no licensor has confirmed it in the record.

Can a company license its data to Stability AI?

No Stability AI program, application page or terms for company records appear in the sources checked as of 2026-11-09.

What did the UK High Court decide about Stability AI's model?

On 2025-11-04 the High Court held that Stable Diffusion is not an infringing copy and found trade mark infringement in part. Getty's appeal was heard in November 2026 and judgment is reserved.

Does Stability AI still exist?

Yes: STABILITY AI LTD, Companies House 12295325, is listed as active (checked 2026-11-09), and it announced a Series B on 2026-08-25.