
The deal tracker holds 10 entries with Google as the acquiring party, and only the Reddit agreement of 2024-02-22 states training.
Every figure here is a dated statement by Google, its counterparty or a court filing, checked 2026-11-09. This independent reference on AI training data licensing covers Google. In the Spirit Airlines auction Google LLC is the successful bidder at USD 10,000,000, and the sale was not approved as of 2026-11-09.
Each number carries its source and check date, and how DataPacts sources and dates every figure is set out in the methodology.
Google's Routes to Training Data: What the Record Shows
Five routes appear in Google's own posts, its counterparties' releases and court filings.
- Platform API licenses, such as Reddit's (2024-02-22)
- News content for AI features, in the News AI pilot (2025-12-10)
- Gemini grounding feeds, such as the Associated Press feed (2025-01-15)
- A bankruptcy bid for Spirit Airlines' data
- Crawling under a robots.txt token
Google documents Google-Extended as a standalone robots.txt product token that controls whether content helps train future Gemini models and grounding, and says it does not affect inclusion in Search or ranking (checked 2026-11-09).
Google's post of 2025-12-10 counts "over 3,000" commercial partnerships with publications, platforms and content providers in more than 50 countries, without stating how many allow training. The tracker lists only those with a public release.
Google Company Identity
The licensing party on record is Google LLC, a Delaware limited liability company and subsidiary of Alphabet Inc., as the Spirit Airlines Bill of Sale shows. The identity table lists each attribute with its source, as of 2026-11-09.
| Attribute | Value | Source |
|---|---|---|
| Legal name | Google LLC | Dkt. 1463, Bankr. S.D.N.Y. 25-11897 |
| Legal form | Delaware limited liability company, per the Bill of Sale | Dkt. 1463 |
| Parent | Alphabet Inc., SEC CIK 0001652044 | SEC EDGAR |
| Website | https://www.google.com | google.com |
| Headquarters country | United States | SEC EDGAR |
| Founded | Not recorded | n/a |
| Type | Frontier lab; developer of Gemini | Google crawler documentation |
| Other entity on record | Google Ireland Ltd., a party in CJEU C-250/25; registry number not recorded | CJEU press release 155/26 |
No founding date and no Irish registry number are stated, because no filing or history page was checked for them.
Google Program Terms and Payouts, Dated
Google documents one paid content program, the News AI pilot for news publications; no program through which a company can offer its own data appears in the sources checked. The program terms table shows each term with the value or record found, as of 2026-11-09.
| Term | What Google 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 |
| News AI pilot eligibility | News publications; how publications join is not stated in Google's posts of 2025-12-10 and 2026-10-15 | News AI pilot | 2026-11-16 | blog.google |
| Pilot payment | "we pay participating news publications for the delivery and use of enhanced content to develop new AI features"; amounts not published | News AI pilot | 2026-11-09 | blog.google |
| Pilot use | Display with attribution and links in Google News AI features; training not stated | News AI pilot | 2026-11-09 | blog.google |
| Pilot scale | More than 400 publications as of 2026-10-15 | News AI pilot | 2026-11-09 | blog.google |
| Commercial agreements overall | "over 3,000" in more than 50 countries (Google, 2025-12-10) | Google claim | 2026-11-09 | blog.google |
| Price in the one company-data bid | Successful bid, a court figure and not a payout; amount in the Spirit section below | Spirit Airlines auction | 2026-11-09 | Dkt. 1463 |
| Payment trigger in that bid | Payment contemporaneous with delivery, subject to court approval | Bill of Sale form | 2026-11-09 | Dkt. 1463 |
| Referral fee | None recorded | n/a | 2026-11-09 | n/a |
| Crawler control | Google-Extended product token | Crawler documentation | 2026-11-09 | developers.google.com |
Google publishes no pilot amount, so no payout can be quoted. Against the program terms compared across buyers that publish ranges, referral fees or license scope, Google's row holds statements without figures. The one price on record, the Spirit bid, is an auction result in a court filing and not a payment by Google.
Reddit's Form S-1 of 2024-02-22 gives one aggregate value for January 2024 licenses, USD 203.0 million, with counterparties not named, so it is not a Google figure.
Google's Licenses and Bid on Record: 10 Entries by Date
The deal tracker holds 10 entries with Google as the acquiring party: nine licenses from Reddit's of 2024-02-22 to the Financial Times' pilot entry of 2026-10-15, and one bankruptcy bid awaiting approval.
The table lists them newest first, as of 2026-11-09, with values as stated by the parties or the court record. The use column reads T for training, D for display or grounding and n/s for not stated.
| Date | Counterparty | What is licensed or bid for | Use | Value | Source |
|---|---|---|---|---|---|
| 2026-10-15 | The Financial Times Limited | Enhanced news content for AI features in Google News (News AI pilot) | D | Not disclosed; Google states it pays participating publications | blog.google |
| 2026-10-12 | Reach plc | Enhanced content rights across its UK and Ireland portfolio (News AI pilot) | D | Not disclosed; Reach quotes Google: "we pay for enhanced content rights" | reachplc.com |
| 2026-08-14 | Spirit Airlines, LLC and affiliated debtors | De-identified business data; passenger profiles not included | n/s | USD 10,000,000 successful bid; no sale order on the docket as of 2026-11-18 | Dkt. 1463 |
| 2026-01-15 | Wikimedia, LLC (Wikimedia Enterprise) | Wikipedia content via the paid Enterprise API; Google listed as an existing partner | n/s | Not disclosed | enterprise.wikimedia.com |
| 2025-12-10 | El País (Prisa Media) | News AI pilot | D | Not disclosed; as above | blog.google |
| 2025-12-10 | Guardian Media Group plc | News AI pilot | D | As above | blog.google |
| 2025-12-10 | The Washington Post (WP Company LLC) | News AI pilot | D | As above | blog.google |
| 2025-01-15 | The Associated Press | Real-time feed to enhance results in the Gemini app | D | Not disclosed | blog.google |
| 2024-02-29 | Stack Exchange, Inc. (Stack Overflow), with Google Cloud | Technical Q&A surfaced in the Google Cloud console; Gemini for Google Cloud integration | n/s | Not disclosed | stackoverflow.co |
| 2024-02-22 | Reddit, Inc. | Data API access "to display, train on, and otherwise use" Reddit content | T | Not disclosed by the parties | blog.google |
Codes follow each party's wording, and n/s does not mean training is excluded.
By use stated, one entry states training (Reddit), six state display or grounding (the Associated Press, the Guardian, the Washington Post, El País, Reach and the Financial Times) and three do not say (Stack Overflow, Wikimedia Enterprise and Spirit). Google's announcement says the Reddit access does not change its use of publicly crawlable content.
No entry carries a value stated by Google or a counterparty. Reddit's 10-Q for the third quarter of 2026 says that agreements with its two main partners are approaching the end of their respective current terms, and the partners are not named in that sentence.
The pilot has more than 400 publications as of 2026-10-15. Participants and terms of the Google News AI pilot program are kept on a separate page, and the amounts stay unpublished.
What Google Agreed to as Buyer in the Spirit Airlines Auction
Google LLC made the successful bid of USD 10,000,000 at the auction of Spirit Airlines' de-identified business data on 2026-08-14, with Mercor.io Corporation as alternate bidder at USD 7,500,000 (Notice of Auction Results, Bankr.
S.D.N.Y. No. 25-11897 (SHL), Dkt. 1463); the sale hearing was set for 2026-11-17 (ECF 1828), and no sale order was on the docket as of 2026-11-18.
The revised schedule in the Notice of Filing of Revised Proposed Sale Order (ECF 1802, 2026-11-09) includes 1,986,737,423 emails across 90,967 accounts, 663,530,348 Teams items, 90,847,707 SharePoint and OneDrive items, 516 repositories with about 30 million lines of code and 175,658 employee records, and excludes 97,500,000 passenger profiles.
The revised sale papers place duties on the buyer. Tonic AI, Inc., the Deidentification Agent, de-identifies the data to the California Consumer Privacy Act standard (45 C.F.R. 164.514 for health data), and Google bears all costs of that process without a price reduction (ECF 1802).
The revised proposed order bars Google from intentionally or knowingly re-identifying the data, and transferees must be bound alike. Payment is contemporaneous with delivery, the sale is as-is and subject to court approval, and New York law governs. Full terms sit with the Spirit Airlines' data archive sale page.
11 U.S.C. 363(b)(1) allows a sale of personally identifiable information covered by a debtor's privacy policy only if consistent with it, or after a consumer privacy ombudsman (11 U.S.C. 332), notice, hearing and court approval.
The ombudsman writes that "from a privacy perspective, approval of the sale of the deidentified data would be appropriate", and notes that risks to employees and third parties were not investigated. micro1 filed a notice of intent to submit a competing bid (Dkt. 1556, 2026-09-03) without an amount.
The flight attendants' union AFA-CWA objected on employee-privacy grounds and states that a letter of 2026-10-08 to the chief executives of Google and Spirit was signed by 121 members of Congress, a count that press reports state differently. No sale order was on the docket as of 2026-11-18.
Google Reputation: Lawsuits and Regulators on Training Data
Four court proceedings and one regulatory inquiry on record concern how Google obtained or used data to train its models. The table lists them by filing date, with status as of 2026-11-09.
| Date | Proceeding | Forum and docket | What it concerns | Status as of 2026-11-09 | Source |
|---|---|---|---|---|---|
| 2023 | In re Google Cloud Contact Center AI Privacy Litigation | N.D. Cal. 3:23-cv-05437-RFL | Call recordings allegedly usable to improve AI models (CIPA) | Orders of 2026-10-22: three plaintiffs to arbitration, three dismissed with leave to amend by 2026-11-12 (Dkt. 140 and 141) | govinfo |
| 2023 | In re Google Generative AI Copyright Litigation | N.D. Cal. 5:23-cv-03440 | Training on books and images | Pending; publishers intervened | govinfo |
| 2024-09-12 | Data Protection Commission inquiry (PaLM 2) | Ireland, s.110 Data Protection Act 2018 | Whether a DPIA under Art. 35 GDPR preceded processing EU personal data for PaLM 2 | Inquiry announced; no reference number public | dataprotection.ie |
| 2025-04-03 | Like Company v Google Ireland Limited | CJEU C-250/25 | Gemini training on press publications and chatbot answers | Advocate General opinion 2026-11-03; judgment pending | curia.europa.eu |
| 2026-07-09 | EVOX Productions, LLC v. Google LLC | N.D. Cal. 5:26-cv-07013-EKL | Copyright claims over a vehicle image library | Pending | complaint |
None of the five has produced a final ruling on whether training on the material is lawful.
In the CJEU case the Advocate General's opinion of 2026-11-03 does not bind the Court; per the press release it treats the training questions as inadmissible because hypothetical and says, in passing, that training large language models is as a rule permitted under the text and data mining exception.
A pending case is an allegation, not a finding. The tracker for lawsuits over training data holds the rulings and settlements of all defendants.
Benefits, Risks, Downsides and Alternatives to Licensing Data to Google
Licensing to Google puts content with a developer that names its counterparties and states that it pays pilot publications, but no Google terms for companies were found and no amount has been disclosed.
Benefits
- Counterparties are named in Google's own posts, as with Reddit, the Associated Press and the pilot publishers
- Google states that it pays participating news publications
- The Spirit form of Bill of Sale is a public contract text with the buyer's duties
Risks
- Litigation and regulatory proceedings over training data are pending, as the table above shows
- A court-supervised bid can fail or change before approval, as the Spirit sale, with no sale order on the docket as of 2026-11-18, shows
- Employees and unions object to the Spirit sale
Downsides
- No amount is disclosed for any entry, so no price can be benchmarked
- Use for training is stated in one of 10 entries
- Google-Extended affects future use only
Alternatives
- Keep the content and set a Google-Extended rule in robots.txt
- Sell 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, with the News AI pilot, whose entry route Google does not state, as the only recorded Google route for publishers.
Google Among Frontier Labs That Acquire Data
Google is one of the frontier labs profiled as data buyers, and its Reddit agreement is the one in its record that states training rights. It is compared with other frontier AI labs as data buyers, among them Anthropic, Meta, Microsoft and OpenAI. 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: most Google counterparties
- US bankruptcy data sales: how a sale reaches a bidder like Google
- How AI companies get training data: the five acquisition routes
- Opting out of AI training: user settings
More About Google as a Data Buyer
Google Workspace Data and Google's Own Training Commitment
Google states that Workspace does not use customer data to train models without the customer's prior permission or instruction, and its API policy bars apps from training AI models on user data (checked 2026-11-09). A company's own Workspace license and export are covered under Google Workspace exports.
Creators, YouTube and Contributor Pay
Google documents no paid contributor or company data program in the sources checked; its YouTube third-party training setting lets creators permit other AI companies to train on their videos, is off by default, and YouTube states it is not facilitating payments.
That setting is a creator permission, not a Google program, and creators selling video for AI training have other routes. Paid contributor work is compared under AI training jobs and platforms.
Where Each Google Case and License Is Tracked
Each record above has its own tracker page with full terms, filings and status.
- In re Google Generative AI Copyright Litigation
- Google Cloud Contact Center AI privacy litigation
- Like Company v. Google
- EVOX Productions lawsuits
- The Associated Press AI licensing deals
- Stack Overflow AI licensing deals
- The Guardian AI licensing deals
- The Washington Post AI licensing deals
- Prisa Media AI licensing deals (El País)
- Financial Times AI licensing deals
- Reach plc AI licensing deals
- Wikimedia Enterprise AI deals
Frequently Asked Questions About Google and Data Licensing
Does Google pay for AI training data?
Google states that it pays news publications in its News AI pilot, but neither Google nor any counterparty has disclosed an amount. Reddit's deal value was never stated by either party.
Can a company license its data directly to Google?
No Google application page or terms for companies appear in the sources checked. Its one company-data acquisition on record is a bankruptcy bid for Spirit Airlines' de-identified records.
Did Google buy Spirit Airlines' data?
No sale had been approved as of November 9, 2026. Google LLC is the successful bidder at the auction, and the approval hearing is set for November 17, 2026.
Is Google using my data to train AI?
Google states that Workspace does not train models on customer data without prior permission or instruction. For Gemini Apps, Google says Temporary Chats are kept for up to 72 hours; the opt-out settings are covered on the opt-out page.