V1 Beta. This is a first release for early review by internal and external audiences.

Skip to content

Source

Reusers of Wikimedia projects' content must, under all circumstances, explicitly identify the source of the information being surfaced in their context. Clear sourcing is the most fundamental attribution component in an information ecosystem where third-party platforms mix and recontextualize Wikimedia content. This recognition supports trust, transparency, and compliance with open licenses. Identifying sources ensures that the contributions of Wikimedia volunteers remain visible and that end users can recognize the origin of the knowledge they are consuming.

We recommend indicating sources using text, ideally hyperlinked to the original work. In audio formats, sources can be verbalized in a manner appropriate to the medium (e.g., "from Wikidata" or "According to Wikipedia..."). When facing space constraints or in interfaces with high information density, it’s acceptable to use the project’s Brand mark to represent the source.

Data sources

This section identifies where and how the data behind this signal can be obtained. It lists the relevant Wikimedia APIs, datasets, or metadata fields that reusers can rely on to implement the signal accurately and consistently.

The Wikimedia Attribution API returns source project information in a source_wiki object. This object includes the source wiki’s name, project family, site ID, as well as site and page languages.

Reference documentation can be found in any wiki's REST API sandbox, such as the REST API sandbox on English Wikipedia →

Share feedback on this beta API →

The Wikimedia Enterprise On-demand API returns Source information via a is_part_of field that doesn't require parsing, and a main_entity and additional_entities fields under “URL/identifiers”, that require parsing (See example).

See documentation →

MediaWiki Action API

To retrieve a project’s global license, use the siteinfo query specifying the general property. The lang and sitename fields may then be used to define the source signal. Like with URL parsing, the lang field returns a two character language code which needs to be expanded for the full value (for example, the site information for English Wikipedia).

See documentation →

Implementation guidance

Minimum requirements

  • State the name of the Wikimedia project from which the content was retrieved in a way that is appropriate to the medium (e.g., “Source: Wikidata”, “From Wikimedia Commons” or “According to Wikipedia…”).

  • The source signal must remain accessible to all users in the context of data reuse, either in visual or auditory form. In visual interfaces, place the source label consistently near the reused content. Flexible placement of this signal is permitted only if strictly necessary (e.g., in a sources panel), as long as it is easily findable, identifiable, and easy to associate with the content presented to end users.

  • Refrain from attributing aggregated content to a single Wikimedia project. Instead, clearly list all the sources being used and, if necessary, include a Modification disclaimer.

Best practices

  • Pair the source label with a visual Brand mark to support the quick identification of Wikimedia projects.

  • A project's Brand mark might be used as a sole indicator of source only when space is limited or in interfaces with a high information density.

Reuse scenarios

Required

Clearly indicating the Wikimedia project on search results pages helps users understand where the information originates and reinforces Wikimedia’s role as a trusted knowledge source.

Learn more about attribution in this context →

AI assistants

Required

In conversational interfaces, explicitly crediting Wikimedia projects clarifies that responses draw on a human-curated source.

Learn more about attribution in this context →