On this page
Local CSV citation wrapper
The examples below use record IDs and public source URLs only. They do not invent or retype buyer or seller names outside the record’s own fields. The blocks are not the tracker’s own citation, CSV and JSON downloads (/deal-tracker.csv, /deal-tracker.json) or a new dataset export.
Save the following as public-deal-citation.csv. This is a local wrapper with its own columns, not the columns of the tracker's own CSV download. The current Deal Tracker dataset snapshot has updated: 2026-10-10; replace that value only after reading the current dataset's updated field. Blank fields mean “read the current tracker row,” not “unknown value equals zero.”
record_id,tracker_url,dataset_updated,source_1_date,source_1_url,source_2_date,source_2_url,schema_fields_to_copy,read_from_tracker,notes
2024-02-google-reddit,https://rightmark.ai/deal-tracker/,2026-10-10,2024-02-23,https://fortune.com/2024/02/23/reddit-60m-deal-google-search-giant-train-ai-models-on-posts/,2024-02-22,https://techcrunch.com/2024/02/22/reddit-says-its-made-203m-so-far-licensing-its-data/,"id;date;buyer;buyers;seller;sellerType;dataType;dataDetail;dealType;status;valueUsd;valueText;valueType;term;notes;sources;indexRow",true,Read current PublicDealSchema values before copying
2023-07-openai-shutterstock,https://rightmark.ai/deal-tracker/,2026-10-10,2023-07-11,https://investor.shutterstock.com/news-releases/news-release-details/shutterstock-expands-partnership-openai-signs-new-six-year,2023-07-13,https://venturebeat.com/ai/shutterstock-signs-6-year-training-data-agreement-with-openai,"id;date;buyer;buyers;seller;sellerType;dataType;dataDetail;dealType;status;valueUsd;valueText;valueType;term;notes;sources;indexRow",true,Read current PublicDealSchema values before copyingThe URLs are source links from Deal Tracker records. Do not treat the URL slug or a headline as a substitute for the tracker row. The local dataset_updated column records the value of dataset.updated when the wrapper was prepared.
Local JSON citation wrapper
{
"template": "citation-kit-1",
"trackerUrl": "https://rightmark.ai/deal-tracker/",
"dataset": {
"updated": "2026-10-10"
},
"records": [
{
"id": "2024-02-google-reddit",
"sourceUrls": [
"https://fortune.com/2024/02/23/reddit-60m-deal-google-search-giant-train-ai-models-on-posts/",
"https://techcrunch.com/2024/02/22/reddit-says-its-made-203m-so-far-licensing-its-data/"
],
"sourceDates": [
"2024-02-23",
"2024-02-22"
],
"copyFieldsFromPublicDealSchema": [
"id",
"date",
"buyer",
"buyers",
"seller",
"sellerType",
"dataType",
"dataDetail",
"dealType",
"status",
"valueUsd",
"valueText",
"valueType",
"term",
"notes",
"sources",
"indexRow"
],
"doNotInfer": [
"pricePerUnit",
"completion",
"demand"
]
}
]
}The JSON keys in this block are fields of the local wrapper. copyFieldsFromPublicDealSchema is a checklist, not a field of the Deal Tracker’s JSON download. The names in that list are the fields of a Deal Tracker record; each sources entry uses the record’s title, publisher, date, and url shape.
Citation rules
- Keep one row per record ID. If two articles describe one transaction, add both URLs to the same record.
- Copy
statusexactly as the current row provides it. Do not normalize or rewrite a pending status. - Copy
valueTypeandvalueTexttogether. Never convert an undisclosed value into zero or an estimate. - Keep unit prices, non-USD values, and undisclosed terms in their source form. Do not calculate a rate from a total unless the source states the unit and method.
- Cite each source publication date and the current
dataset.updatedvalue. - Do not add private Rightmark transactions, unpublished buyer identity, or confidential terms.
Record fields to copy
id | date | buyer | buyers | seller | sellerType | dataType | dataDetail | dealType | status | valueUsd | valueText | valueType | term | notes | sources | indexRow
The list is a citation checklist, not a ranking or application export. Read the current row before copying values. Acquisitions, settlements, negotiations, losing bids, and unapproved bids follow the method’s own rules.
Sources
- Rightmark: Deal Tracker
- Fortune: Reddit has struck a $60m deal with Google that lets the search giant train AI models on its posts (opens in a new tab)
- TechCrunch: Reddit says it's made $203M so far licensing its data (opens in a new tab)
- Shutterstock: Shutterstock Expands Partnership with OpenAI, Signs New Six-Year Agreement to Provide High-Quality Training Data (opens in a new tab)
- VentureBeat: Shutterstock signs 6-year training data agreement with OpenAI (opens in a new tab)