Keyboard shortcuts

Press or to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Integrations

Connect to free safety-oriented APIs like Google Content Safety API, OpenAI Moderation API, Zentropi CoPE, and more at the click of a button. We’ve built several integrations into Coop; just enter your API key.

Built-in integrations

For specific integration information and detailed requirements, see that integration’s documentation:

IntegrationCostRequirements
Google Content Safety APIFreeAPI key, approval by Google1
Hasher-Matcher-Actioner (HMA)FreeYour own hashes, and/or access to third-party hash banks2
NCMEC ReportingFreeCyberTip API key, approval by NCMEC3
OpenAI Moderation APIFree4OpenAI API key
Zentropi CoPEFree, paid options5Zentropi API key

Model cards

Each integration includes a model card for consistent, comparable information about how it works. A model card is a short document describing a machine learning model’s intended use, behavior, and limitations; think of it as a nutrition label for AI classifiers.

Each integration’s model card describes:

FieldDescription
PurposeWhat the model is designed to detect or classify
InputThe type of content the model accepts (images, text, URLs, etc.)
OutputThe format and meaning of the model’s response
LimitationsKnown gaps, failure modes, or content types the model handles poorly
RequirementsAccess, approval, or configuration needed to use the integration
Best practicesRecommendations for getting reliable results

No automated classifier is perfect. Model cards help you understand when to trust a signal, what manual review is required, and how to configure rules meaningfully.


  1. Industry and civil society third parties seeking to protect their platform against abuse can apply to access the Content Safety API. Mention in your application that you are using the Coop review tool. Applications are subject to approval and require accepting Google’s terms and conditions.

  2. You don’t need credentials or licenses for using your own hash banks, i.e. if you have your own collection of known violations. Access to each third-party hash requires access from that org; for example, NCMEC needs to provide Hash Sharing API credentials, Tech Against Terrorism provides access to their hash bank.

  3. Requires NCMEC ESP registration and approval to receive CyberTip API credentials

  4. Per OpenAI, the API is completely free and does not count towards monthly usage limits.

  5. Zentropi text classifiers are free, and currently use the openly-licensed CoPE-A-9B model. The API supports all labelers you create, including free and optionally paid; see Zentropi pricing details.