The deal tracker holds 13 entries with Microsoft as the acquiring party: seven state grounding, five state display and one does not state a use. Nine court proceedings name Microsoft or its subsidiary GitHub as a defendant.
Every figure here is a dated statement by Microsoft, its counterparty or a court record, checked 2026-11-09, and no party discloses a value for any entry. This independent reference on AI training data licensing profiles Microsoft by the same method as every other buyer.
Each number carries its source and check date, and how DataPacts sources and dates every figure is set out in the methodology.
Microsoft's Routes to Content and Data: What the Record Shows
Five routes appear in Microsoft's own posts, its subsidiaries' terms and its counterparties' announcements.
- Copilot Daily display licenses with named news sources (2024-10-01)
- The Publisher Content Marketplace, a pilot for grounding (2026-02-03)
- The paid Wikimedia Enterprise API (2026-01-15)
- Data from subsidiaries' own users, at GitHub and LinkedIn
- Gameplay recorded in an own-studio game, used for the Muse model (2025-02-19)
GitHub trains on interaction data from Copilot Free, Pro and Pro+ users from 2026-04-24 unless the user opts out in settings (announced 2026-03-25).
LinkedIn uses member data to train content-generating AI models unless the member turns off the setting "Data for Generative AI Improvement", and the change is not retroactive. Both are the subsidiaries' own terms, not purchases, and they sit under user data and platform terms in the wider picture.
On crawling, no Microsoft AI-training crawler and no robots.txt token for it appear in the record, and no Microsoft statement on the shared AI training opt-out is on record as of 2026-11-10. The European Commission page on the General-Purpose AI Code of Practice (published 2025-07-10, updated 2026-10-07) lists Microsoft among 22 signatories; the Code's Measure 1.3 says signatories employ crawlers that follow robots.txt as specified in IETF RFC 9309, which is the Code's text and no statement about a Microsoft crawler. No Microsoft training-content summary under Article 53(1)(d) of Regulation (EU) 2024/1689 was located.
Microsoft Company Identity
The licensing party in Microsoft's announcements and in court records is Microsoft Corporation, registered with the SEC under CIK 0000789019 with Washington as its state. The identity table lists each attribute with its source, as of 2026-11-09.
| Attribute | Value | Source |
|---|---|---|
| Legal name | Microsoft Corporation | SEC EDGAR |
| SEC registration | CIK 0000789019, "MICROSOFT CORP", Washington; EDGAR checked 2026-11-09 | SEC EDGAR |
| Website | https://www.microsoft.com | microsoft.com |
| Headquarters country | United States | SEC EDGAR |
| Founded | Not recorded | n/a |
| Type | Frontier lab; operator of Copilot and the Publisher Content Marketplace | Microsoft Advertising blog, 2026-02-03 |
| Subsidiary on record | GitHub, Inc.; no own registry number published | GitHub Terms of Service |
| Subsidiary on record | LinkedIn Corporation, and LinkedIn Ireland Unlimited Company for members in the EU, EEA and Switzerland; registry numbers not checked | LinkedIn User Agreement s.1.1 |
No founding date is recorded and no LinkedIn registry number was checked, so neither appears as a fact here.
Microsoft Program Terms and Payouts, Dated
Microsoft documents one content program, the Publisher Content Marketplace for publishers and AI builders; no program through which a company can offer its own records appears in the sources checked. The program terms table shows each term with what Microsoft states or the record shows, as of 2026-11-09.
| Term | What Microsoft states or the record shows | Program or record | Checked | Source |
|---|---|---|---|---|
| Company data program or application page | None recorded | n/a | 2026-11-09 | n/a |
| Marketplace status and process | Pilot; the post of 2026-02-03 links an interest form for publishers and AI builders; no application process described | Publisher Content Marketplace | 2026-11-09 | Microsoft Advertising |
| Marketplace payment | Publishers "will be paid on delivered value", with usage-based reporting; no rates published | Publisher Content Marketplace | 2026-11-09 | Microsoft Advertising |
| Marketplace license | Publishers define licensing and usage terms; AI builders license content for specific grounding scenarios; training not mentioned | Publisher Content Marketplace | 2026-11-09 | Microsoft Advertising |
| Marketplace participants | Co-designed with seven named publishers "among others"; Microsoft states it is "beginning to onboard demand partners, including Yahoo" | Publisher Content Marketplace | 2026-11-09 | Microsoft Advertising |
| Copilot Daily sources | "Copilot Daily will only pull from authorized content sources"; Microsoft's post does not mention payment | Copilot Daily | 2026-11-09 | blogs.microsoft.com |
| Referral fee | None recorded | n/a | 2026-11-09 | n/a |
| Customer data in Microsoft 365 Copilot | Prompts, responses and data accessed through Microsoft Graph are not used to train foundation LLMs (Microsoft's claim) | Microsoft 365 Copilot | 2026-11-09 | learn.microsoft.com |
| Crawler control | No Microsoft AI-training crawler or robots.txt token in the record | n/a | 2026-11-09 | n/a |
Microsoft states that marketplace publishers are paid on delivered value and publishes no rate, so no payout can be quoted; against the program terms compared across buyers that publish ranges or referral terms, Microsoft's row holds statements without figures. The Copilot Daily post names the sources but says nothing about payment to them.
For companies the record is an absence: no application page, no referral fee and no license terms for company records appear in the sources checked.
Microsoft's Content Licenses on Record: 13 Entries by Date
The deal tracker holds 13 entries with Microsoft as the acquiring party: five Copilot Daily content sources named on 2024-10-01, Wikimedia Enterprise's paid API on 2026-01-15, and seven Publisher Content Marketplace co-design publishers named on 2026-02-03. The table lists them newest first, as of 2026-11-09, with values as stated by the parties.
The use column reads G for grounding, D for display and n/s for not stated.
| Date | Counterparty | What is licensed | Use | Value | Source |
|---|---|---|---|---|---|
| 2026-02-03 | The Associated Press | Premium publisher content for grounding AI answers, Copilot first (marketplace co-design publisher) | G | Not disclosed; Microsoft states publishers "will be paid on delivered value" | Microsoft Advertising |
| 2026-02-03 | Business Insider Inc (Axel Springer brand) | As above | G | As above | As above |
| 2026-02-03 | Condé Nast | As above | G | As above | As above |
| 2026-02-03 | Hearst Magazines | As above | G | As above | As above |
| 2026-02-03 | People Inc | As above | G | As above | As above |
| 2026-02-03 | USA TODAY Co. | As above | G | As above | As above |
| 2026-02-03 | Vox Media LLC | As above | G | As above | As above |
| 2026-01-15 | Wikimedia, LLC (Wikimedia Enterprise) | Wikipedia and Wikimedia content via the paid Enterprise API; Microsoft named among new AI customers | n/s | Not disclosed | enterprise.wikimedia.com |
| 2024-10-01 | Reuters (Thomson Reuters) | Authorized news content for Copilot Daily | D | Not disclosed; Microsoft's post does not mention payment | blogs.microsoft.com |
| 2024-10-01 | Axel Springer SE | As above | D | As above | As above |
| 2024-10-01 | Hearst Magazines | As above | D | As above | As above |
| 2024-10-01 | USA TODAY Network (Gannett Co., Inc.; now USA TODAY Co., Inc.) | As above | D | As above | As above |
| 2024-10-01 | The Financial Times Limited | As above | D | As above | As above |
Codes follow each party's wording, and n/s does not mean training is excluded. HarperCollins and Taylor & Francis are not rows, as the next section explains.
By use stated, seven entries state grounding, five state display in Copilot Daily and one does not say; none states training. No entry carries a value stated by Microsoft or a counterparty.
Microsoft's own wording on the marketplace is that publishers define their terms and are paid on delivered value, and its post does not mention training. Participants, process and terms of the Microsoft Publisher Content Marketplace are kept on a separate page.
Book and Journal Licenses No Party Attributes to Microsoft
Two 2024 licenses of books and academic content appear in search results under Microsoft's name, but neither licensor nor any court record names Microsoft as the licensee.
For HarperCollins, the Authors Guild describes a November 2024 license of nonfiction backlist titles to "an unknown AI company", with a separate opt-in addendum per author and USD 5,000 per title split 50-50 between author and publisher (Authors Guild, 2024-11-19).
For Taylor & Francis, Informa names a "major AI/Large Language Model buyer" in its May 2024 AGM update and a "second major partnership secured" in its half-year report of 2024-07-24, without naming either buyer.
DataPacts lists a deal under a buyer only when a party or a court names that buyer, so neither license is a row in the table above.
Microsoft Reputation: Court Proceedings Naming Microsoft or GitHub
Nine court proceedings on record name Microsoft or its subsidiary GitHub as a defendant over the use of works to build AI models or over model outputs; eight of them sit in, or are conditionally transferred to, MDL No. 3143 in the Southern District of New York. The table lists them by docket, with status as of 2026-11-09.
| Proceeding | Court and docket | Microsoft's role | What it concerns | Status as of 2026-11-09 | Source |
|---|---|---|---|---|---|
| Doe 1 et al. v. GitHub, Inc. et al. | N.D. Cal. 4:22-cv-06823-JST; 9th Cir. No. 24-7700 | Defendant with GitHub and OpenAI entities | DMCA 1202(b) claims over Copilot and Codex output | Ninth Circuit affirmed dismissal of the DMCA claim on 2026-09-16; mandate 2026-10-08 (Dkt. 297); back in district court | ca9.uscourts.gov |
| Authors Guild et al. v. OpenAI Inc. et al. | S.D.N.Y. 1:23-cv-08292-SHS-OTW (MDL 3143) | Defendant | Books used to train GPT models | Summary-judgment motions briefed; no ruling located | jpml.uscourts.gov |
| Alter et al. v. OpenAI, Inc. et al. | S.D.N.Y. 1:23-cv-10211-SHS-OTW (MDL 3143) | Defendant | Books used to train GPT models | Pending in the MDL | As above |
| The New York Times Company v. Microsoft Corporation et al. | S.D.N.Y. 1:23-cv-11195-SHS-OTW, filed 2023-12-27 (MDL 3143) | First-named defendant | Copyright, DMCA 1202, misappropriation, trademark dilution | Summary-judgment stage; no ruling located | Dkt. 514 |
| Basbanes et al. v. Microsoft Corporation et al. | S.D.N.Y. 1:24-cv-00084 (MDL 3143) | First-named defendant | Training on books | Pending in the MDL | jpml.uscourts.gov |
| The Intercept Media, Inc. v. OpenAI, Inc. et al. | S.D.N.Y. 1:24-cv-01515 (MDL 3143) | Claims against Microsoft dismissed 2025-02-20 | DMCA 1202(b) removal of copyright management information from articles used in training | Pending against OpenAI in the MDL | Dkt. 514 |
| Daily News LP et al. v. Microsoft Corporation et al. | S.D.N.Y. 1:24-cv-03285-SHS, filed 2024-04-30 (MDL 3143) | First-named defendant | Same claims as The Times | Summary-judgment stage | As above |
| The Center for Investigative Reporting, Inc. v. OpenAI, Inc. et al. | S.D.N.Y. 1:24-cv-04872 (MDL 3143) | Defendant | Copyright and DMCA claims over news content | Pending in the MDL | jpml.uscourts.gov |
| Emmerich Newspapers, Inc. et al. v. Microsoft Corporation et al. | S.D. Miss. 3:26-cv-00791 | First-named defendant | Claims not recorded | Conditional transfer order CTO-4 entered 2026-10-27 (MDL 3143 Dkt. 107) | jpml docket |
For Microsoft as a party, the opinion of 2025-04-04 (Dkt. 514) lets the contributory-infringement claims and the older direct-infringement claims survive, dismisses misappropriation with prejudice and dismisses the section 1202(b)(1) claims against Microsoft in all three news actions, and the Times' motion for leave to amend its contributory claim against Microsoft was denied on 2026-08-06. None of the nine has produced a final ruling on whether training on the material is lawful, and a pending case is an allegation, not a finding.
Claims, rulings and status of The New York Times v. Microsoft and OpenAI are kept on its own page, and the member actions share the MDL page listed below.
Benefits, Risks, Downsides and Alternatives to Licensing Content to Microsoft
Licensing to Microsoft puts content with a developer that names its content sources and states that marketplace publishers are paid on delivered value, but no rate, amount or company-data terms have been published.
Benefits
- Counterparties are named in Microsoft's own posts, as with the seven marketplace publishers and the five Copilot Daily sources
- Microsoft states that publishers define their own licensing and usage terms
- Microsoft states that payment follows delivered value, with usage-based reporting
Risks
- Court proceedings over training data name Microsoft in nine matters, as the table above shows
- GitHub (for Copilot Free, Pro and Pro+ users) and LinkedIn train on their own users' data by default unless users opt out
- A pilot's terms can change before it leaves the pilot stage
Downsides
- No payment rate or amount is published for any entry
- Use is stated as grounding or display in 12 of 13 entries and is not stated in the 13th
- No Microsoft crawler token is on record, so there is no robots.txt control to point to
Alternatives
- Keep the content and apply the controls the shared AI training opt-out page describes
- License to a different buyer type, such as a data vendor or marketplace
- Not license at all, which is a valid outcome
The alternatives to selling data include keeping it, since Microsoft documents no route for company records.
Microsoft Among Frontier Labs That Acquire Data
Microsoft is one of the frontier labs profiled as data buyers, and the only one in the record whose publisher program is described as a marketplace with demand partners. It is compared with other frontier AI labs as data buyers, among them Amazon, Google, Meta and OpenAI, its co-defendant in MDL No. 3143. Related pages:
- AI data licensing deals: the tracker of all agreements
- Who buys AI training data: every buyer and program compared
- Choosing a buyer for your data: a lab is one of several routes
- News publishers licensing content to AI companies: all 13 counterparties are publishers or content platforms
- Book publishers and authors licensing books: how books are licensed, buyer named or not
- Opting out of AI training: the GitHub and LinkedIn settings
- Lawsuits over training data: every case with rulings
More About Microsoft as a Data Buyer
Data a Company Holds in Microsoft 365, Teams and GitHub
Microsoft states that prompts, responses and data accessed through Microsoft Graph in Microsoft 365 Copilot are not used to train foundation LLMs (checked 2026-11-09), so a company's own Microsoft 365 records stay its own to license or keep.
- Microsoft 365 exports for data licensing
- Microsoft Teams data: export and eDiscovery
- GitHub repositories and their export rights
Gameplay Recordings, Contributors and Paid Work
Microsoft states that its Muse model trained on more than seven years of continuous human gameplay from Ninja Theory's Bleeding Edge, recorded when the player agreed to the EULA (2025-02-19).
Studios licensing gameplay data for AI are covered separately. No paid contributor program by Microsoft appears in the record, and paid work is compared under AI training jobs and platforms.
Where Each Microsoft Case and License Is Tracked
Each record above has its own tracker page with full terms, filings and status.
- In re OpenAI Copyright Infringement Litigation (MDL 3143)
- Daily News v. Microsoft and OpenAI
- The Intercept v. OpenAI
- Authors v. OpenAI class actions
- Doe v. GitHub (Copilot)
- The Associated Press AI licensing deals
- Axel Springer AI licensing deals (including Business Insider)
- Condé Nast AI licensing deals
- Hearst AI licensing deals
- People Inc. AI deals
- USA TODAY Co. AI licensing deals
- Vox Media AI licensing deals
- Reuters AI licensing deals
- Financial Times AI licensing deals
- Wikimedia Enterprise AI deals
- Microsoft Publisher Content Marketplace and Copilot news deals
- HarperCollins book licensing for AI (licensee not named)
- Taylor & Francis AI licensing deals (licensee not named)
Frequently Asked Questions About Microsoft and Data Licensing
Does Microsoft pay for AI training data?
Microsoft states that publishers in its Publisher Content Marketplace will be paid on delivered value, for grounding rather than training. It has published no rate, and no party has disclosed an amount.
Can a company license its data directly to Microsoft?
No Microsoft application page or terms for company records appear in the sources checked. Its one documented program, the Publisher Content Marketplace, takes interest from publishers and AI builders through a form.
Did Microsoft license HarperCollins books for AI training?
No party has said so. The Authors Guild describes the November 2024 HarperCollins nonfiction license as one with an unknown AI company, paying USD 5,000 per title split 50-50 with opted-in authors (Authors Guild, 2024-11-19).
Does Microsoft train AI on Microsoft 365 data?
Microsoft states that Microsoft 365 Copilot prompts, responses and Graph data are not used to train foundation LLMs (checked 2026-11-09). Its subsidiary GitHub trains on Copilot interaction data of Free, Pro and Pro+ users from 2026-04-24 unless the user opts out.