DataPacts

Stack Overflow AI Licensing Deals: Terms, Value and Sources

Checked

At a glance

Stack Exchange, Inc., the company behind Stack Overflow, announced two AI agreements in 2024, with Google Cloud on 2024-02-29 and with OpenAI on 2024-05-06, and states no value for either.

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 Google Cloud release says Google Cloud "will also surface validated technical knowledge from Stack Overflow directly in the Google Cloud console", and the OpenAI release states that OpenAI uses Stack Overflow's OverflowAPI product and provides attribution in ChatGPT.

Beyond the two agreements, Stack Overflow offers its Q&A corpus to any AI developer through Stack Data Licensing, with attribution expected and no published price (checked 2026-11-09).

This page sets the three side by side for readers of AI training data licensing, with each term credited to the party that stated it and the gaps listed as of 2026-11-17.

Fact-check note: every term on this page is attributed to the party that stated it and dated; the approach is described in how we source, date and check every figure.

Stack Overflow's two announced AI agreements at a glance

Stack Overflow's own press archive holds both agreements, and each release names the counterparty, the date and the products involved.

The table compares the two agreements as Stack Overflow states them (as of 2026-11-17).

Date Counterparty What it receives, as stated Use stated Value Stated by Source
2024-02-29 Google Cloud (Google LLC) validated technical knowledge from Stack Overflow surfaced in the Google Cloud console; Gemini for Google Cloud integration not stated not disclosed Stack Overflow press release stackoverflow.co
2024-05-06 OpenAI OverflowAPI access; attribution in ChatGPT; collaboration to improve model performance; Stack Overflow uses OpenAI models for OverflowAI model improvement: OpenAI will "improve its AI models using enhanced content and feedback from the Stack Overflow community" not disclosed Stack Overflow press release stackoverflow.co

Seller in both: Stack Exchange, Inc. (SEC CIK 0001491819, formerly Stack Overflow Internet Services, Inc.). The parent tracker shows both rows with the value "n/d".

Google Cloud, 2024-02-29: Stack Overflow answers in the Google Cloud console

Stack Overflow announced a "strategic generative AI partnership" with Google Cloud on 2024-02-29, under which Google Cloud "will also surface validated technical knowledge from Stack Overflow directly in the Google Cloud console", in Stack Overflow's words.

The data involved is technical Q&A from Stack Overflow, shown in the console and in an integration with Gemini for Google Cloud. The release as recorded names no value, no term and no volume.

Stack Overflow is the stating party; the record holds no separate statement by Google Cloud on the terms, so this page reports only what Stack Overflow says.

The release does not state that Stack Overflow content will train a Google model. That sets it apart from Reddit's announcement of 2024-02-22, in which Google gets Reddit Data API access "to display, train on, and otherwise use" Reddit content.

The Stack Overflow text describes surfacing and integration, and this page adds no use that the text omits.

OpenAI, 2024-05-06: OverflowAPI access and attribution in ChatGPT

Stack Overflow states that OpenAI uses its OverflowAPI product and provides attribution in ChatGPT, and that the two companies collaborate to improve model performance, in a release dated 2024-05-06.

OverflowAPI is Stack Overflow's API product for its technical Q&A. The data in this agreement is that Q&A, delivered through the API, and the release states attribution as a term: OpenAI will "provide attribution to the Stack Overflow community within ChatGPT".

The agreement runs in both directions. Stack Overflow states that it uses OpenAI models for its own OverflowAI product, so Stack Overflow is a buyer of model access as well as a seller of data access.

The release (checked 2026-11-17) adds that the integration will help OpenAI "improve its AI models using enhanced content and feedback from the Stack Overflow community", a model improvement statement without the word training. No value, term, volume or exclusivity term is stated.

Stack Data Licensing: the standing offer to any AI developer

Stack Overflow states, checked 2026-11-09, that it licenses Q&A pairs from 150+ Stack Exchange sites, with tags, comments, votes and revisions, for continuous access in training and fine-tuning, with attribution expected.

The table lists the program's terms as Stack Overflow states them (as of 2026-11-09).

Term As Stack Overflow states it Source
Corpus 83M+ Q&A stackoverflow.co/data-licensing
Scope Q&A pairs from 150+ Stack Exchange sites; tags, comments, votes, revisions stackoverflow.co/data-licensing
Access continuous access for training and fine-tuning stackoverflow.co/data-licensing
Attribution expected stackoverflow.co/data-licensing
Sample 1,000 Q&A pairs stackoverflow.co/data-licensing
Process sample dataset, API documentation, sales contact stackoverflow.co/data-licensing
Price not published stackoverflow.co/data-licensing
Eligibility AI companies stackoverflow.co/data-licensing

The process starts with a free sample of 1,000 Q&A pairs and the API documentation, then a conversation with Stack Overflow's sales contact. A buyer therefore cannot benchmark a price from the page, and a prospective licensee learns terms only after contacting Stack Overflow.

Eligibility is stated as AI companies, so the offer is aimed at developers, not at individual contributors. The same structure of tags, votes and revisions is what makes a forum archive more than raw text, a point developed for forum and community archives as AI training data.

Why no value appears for either Stack Overflow agreement

Neither Stack Overflow nor its counterparties have published a value for either agreement, and both records read "not disclosed".

The record also holds no owner of Stack Exchange, Inc. above the company itself, and no parent-company report of the value of the agreements or of licensing revenue. This page therefore states no figure, not even as a reported one.

Platforms that license user content sometimes disclose licensing income in filings, and what those filings show is covered for platforms and online communities licensing user content. The page here adds a figure only if a party or a filing states it, and then with the stating party and the date.

Attribution and the content license: what Stack Overflow has stated on record

Attribution is the one term both of Stack Overflow's AI texts on record share: the OpenAI release places attribution in ChatGPT, and the Stack Data Licensing page expects attribution from licensees.

The two statements differ in kind. The first describes what a product shows to its users, and the second describes what a licensee is expected to do. Neither states how attribution is displayed, measured or enforced.

This page holds no Stack Overflow statement on how contributors responded to the agreements, and none on the license that governs user posts. Those questions are left open here rather than answered from press coverage.

Creative Commons states that its licenses apply only where copyright permission is required (checked 2026-11-09), and that content under no-derivatives terms should not be used as training data on a cautious reading.

How such licenses interact with AI training is covered for dataset licenses for AI training such as CC. This page states no view on whether training triggers any license condition. It is information, not legal advice.

What the public record does not show about these agreements

As of 2026-11-17, no source on record states the value, the term, any exclusivity or the volume delivered of either agreement, any training use under the Google Cloud agreement, nor who owns Stack Exchange, Inc.

  • Value: no party states what either counterparty provides in return.
  • Term: no start of delivery, length or renewal is stated.
  • Exclusivity: no source states an exclusivity term or its absence.
  • Volume: the amount of data delivered through OverflowAPI is not stated.
  • Training use: the Google Cloud release states no training use; the OpenAI release states model improvement, not training in those words.
  • Ownership: no owner of Stack Exchange, Inc. above the company is on record.

A statement by Stack Overflow, Google or OpenAI, a parent-company report or a filing would close these gaps, and the page is updated when one appears.

What Stack Overflow's record shows a holder weighing an API license

The record shows how one community platform licenses ongoing API access, not what another holder will be offered.

Benefits

  • Continuous access keeps a standing relationship with a buyer, and attribution is a stated term.
  • Two named counterparties on record show that more than one AI developer takes API access to the same corpus.

Risks

  • The record holds no Stack Overflow statement on contributor response or on the license of user posts, so a holder cannot read those effects from it.
  • No exclusivity term is stated, so what a counterparty may do with the same corpus is unknown.

Downsides

  • No published price exists to benchmark against.
  • No term is stated, so the length of access is unknown.

Alternatives

  • A one-off file instead of an API.
  • Licensing under open terms only.
  • Not licensing at all.

The weighing of these routes for a holder is set out under whether to sell or license data to AI companies.

Where to read the deal tracker, code and platform pages around Stack Overflow's agreements

This page covers Stack Overflow's agreements only; the pages below cover every recorded deal, the buyers, other platform licenses and company code.

A company's own codebase, made of repositories rather than public Q&A, is licensed on different terms and covered on the page for selling or licensing source code to AI companies.

Related pages on Automattic (Tumblr, WordPress.com) data licensing, Yelp and OpenAI data licensing agreement and training data attribution and per-use payment models follow in later batches.

Frequently asked questions

Did Stack Overflow license its data to OpenAI?

Yes, Stack Overflow states that OpenAI uses its OverflowAPI product and provides attribution in ChatGPT, announced 2024-05-06. The release also describes a collaboration to improve model performance.

How much did OpenAI or Google pay Stack Overflow?

No party has disclosed a value for either agreement. The records of 2024-02-29 and 2024-05-06 read "not disclosed". Readers who want new disclosures as they appear can follow the deal tracker updates by email.

Does Google train its models on Stack Overflow content?

The 2024-02-29 announcement as recorded states surfacing in the Google Cloud console and a Gemini for Google Cloud integration, not training. No source on record states a training use.

What is OverflowAPI?

OverflowAPI is Stack Overflow's API product for licensing its Q&A data. OpenAI uses it under the agreement announced 2024-05-06.