Legal:Wikimedia Foundation Data Retention Guidelines

Introduction

Data is important. It is one of the ways we can learn and grow as an organization and a movement, and how we can help make the projects better for those who use them to create, learn, and share. At the same time, we are committed to keeping your personal data "for the shortest possible time that is consistent with the maintenance, understanding, and improvement of the Wikimedia Sites, and our obligations under applicable U.S. law" (quote from the Wikimedia Foundation Privacy Policy).

This document helps explain how we fulfill this commitment, by describing our guidelines for data retention, system design, and ongoing auditing and maintenance. These guidelines are meant to be a living document — they will be updated over time to reflect current retention practices.

To what data do these guidelines apply?

These guidelines apply to all non-public data we collect from Wikimedia Sites covered by the Privacy Policy and Non-Wiki Privacy Policy. Our Donor Privacy Policy includes separate data retention guidelines that apply to donor information.

How long do we retain non-public data?

Unless otherwise indicated, we retain the following types of data for no more than the following periods of time:


Data type	Origin	Examples	Maximum Retention Period
Non Public Personal information	Collected automatically from a user	IP addresses of site visitors (operational data) IP addresses of A/B test subjects (analytical data) Identifying user-agent information of site visitors	After at most 90 days, it will be deleted, aggregated, or de-identified
Non Public Personal information	Account settings	Email address	Until user deletes/changes the account setting.
Non-personal information	Collected automatically from a user	Data collected by MediaWiki about a user account's activity (e.g., first time a user goes to an edit page, date and time that a user verifies their email address)	Indefinitely
	Collected automatically from a user	Data collected by EventLogging and associated with their user ID (e.g., whether an account was created on mobile, A/B test data for Getting Started)	After at most 90 days, it will be deleted, aggregated, or de-identified
	Provided by a user	Logs of terms entered into the site's search box, or terms within prefilled links to the search engine that have been followed by user navigation	After at most 90 days, it will be deleted, aggregated, or de-identified
	Provided by a user	Language	Until user deletes/changes the account setting.
Non-personal information not associated with a user account^{[T 1]}	Collected automatically from various users	Counts of how many times certain events have occurred (e.g. successful HTTPS requests)	Indefinitely
Articles browsed by readers	Collected automatically from a reader	A list of articles visited by readers	After at most 90 days, if retained at all, then only in aggregate form

↑ For the purposes of this table, "user account" means username, user ID, or IP address; "reader" means visitor to a Wikimedia project.

How long do we retain public data?

Wikimedia hosts Wikipedia and the associated projects as part of our mission to collect, document, and freely distribute the sum of human knowledge to the world. Accordingly, when you make a contribution to any Wikimedia Site, including on user or discussion pages, you are creating a permanent, public record of every piece of content added, removed, or altered by you. The page history will show when your contribution or deletion was made, as well as your username (if you are signed in) or your IP address (if you are not signed in). We may use your public contributions, either aggregated with the public contributions of others or individually, to create new features or data-related products for you, or to learn more about how the Wikimedia Sites are used. If you mistakenly included your personal information in a contribution to a Wikimedia Site and you would like to have it removed, please consult the community's oversight policy. Keep in mind that the transparency and integrity of our sites' revision histories is essential to our mission, and the Foundation supports our community's right to reject oversight requests in order to protect the projects.

If you choose to register for an account with the Wikimedia projects, you will be asked to select a username. Usernames are retained until the user requests that the account be renamed, or goes through the community courtesy vanishing process.

For more information, see our Privacy Policy.

Definitions

For the purposes of these guidelines:

"Personal information" means information you provide us or information we collect from you that identifies or could be used to personally identify you. For details, please see the Wikimedia Foundation Privacy Policy and Non-Wiki Privacy Policy.
Some examples of "public information" would include:
- (a) your IP address, if you edit without logging in;
- (b) your gender, if it is disclosed under your user profile;
- (c) any personal information you disclose publicly on the Wikimedia Sites, such as your real name or age.
Some examples of types of information that are considered to be "nonpublic information" include:
- (a) your IP address, if you edit while logged in;
- (b) your email address, if you provided one to us during account registration (but did not post it publicly); and
- (c) your general location information as might be derived from your IP address, if you have not posted it publicly. The types of information that are considered "nonpublic" as opposed to "public" are more fully explained in our Privacy Policy.
Data is "de-identified" when it has been aggregated or otherwise retained in a manner such that it can no longer be used to identify the user.
Data is "aggregated" when the data associated with a specific user has been combined with data from others to show general trends or values without identifying specific users.

An example of how data can be aggregated includes:

Using ranges rather than specific numbers, such as recording that there are "between 1 and 10 editors in language X in country Y" rather than recording that there are 4 editors.

Terms that are not defined in this document have the same meaning given to them in the Privacy Policy.

Exceptions to these guidelines

If we make exceptions to these guidelines, we will notify the community by describing the exception on this page.

Data may be retained in system backups for longer periods of time, not to exceed 5 years.
When we conduct a survey or other research, we will provide you with a privacy statement specifying the term of retention for information (including personal information) collected through your participation in such research. In certain cases, information may be retained indefinitely for educational, development, or other related purposes, unless otherwise indicated in the relevant privacy statement. Such information may be retained in raw, aggregated, or de-identified form until we receive a request from the participant to delete the information.
Research related to COVID-19: The Wikimedia Foundation Research team is conducting research regarding COVID-19 and its impact on Wikipedia. Retaining de-identified readership data from COVID-19 related articles will enable us to better understand how to prioritize content creation, to understand what happens to readership when there is a "shock to the system", and to empower the research community to answer such questions. By "COVID-19 related articles", we mean articles that link to the COVID-19, SARS-CoV-2 and 2019-2020 COVID-19 pandemic Wikidata items. For comparison purposes, we will retain data from a small number of articles unrelated to COVID-19 as well. In order to collect sufficient data, and obtain a picture of readership as time passes, we will be retaining this de-identified data beyond the 90-day retention limit, for a period of one year, ending on March 1, 2021. (Note that this includes a one-month extension due to staffing changes, in order to allow for the project's completion.). For technical details about the sampling and de-identification process, please see the project page on GitHub.
Editing research: There is a short-term extension applying to data collected as part of experimental features to improve replying on talk pages. In order to collect and analyze sufficient data, this data must be kept beyond the standard 90-day period. The retained data will be deleted, aggregated, or de-identified within 180 days.
Campaign landing pages: for certain events, campaigns, or marketing channels, users may create accounts on special landing pages. After creating their account on those pages, the association between their account and its source may be retained indefinitely, both to provide a good user experience for that account and for longitudinal analysis on campaign effectiveness. For more information, contact mmillerwikimedia.org.
CampaignEvents extension: An exception exists for data collected by the CampaignEvents extension. The extension collects the global user IDs of event organizers and event participants, as well as which events users organized or attended and when participants registered for an event. In order for the extension features to work consistently, data collected by the CampaignEvents extension may be retained indefinitely.
Sound logo contest: There is a short-term extension applying to data collected as part of contest entries to allow the brand studios team to evaluate entries in preparation for announcing the winner in February 2023. The retained data will be deleted, aggregated or de-identified within 90 days after the winner is announced.
Webrequest datasets: There is a short, one-time extension for data from the wmf.webrequest and wmf.pageview_actor datasets. This data needs to be retained longer than usual while we correct an error in the way unique devices are calculated from the dataset. Accurate unique device statistics are necessary for engineering purposes and legal reporting requirements. The underlying data used to calculate these statistics will be retained for an extra 30 days beyond the ordinary 90-day deletion period. After 30 days, the affected data will be purged and retention settings will reset back to 90 days.
In rare cases, we, or particular users with certain administrative rights as described in our Privacy Policy, may need to retain your personal information, including your IP address and user agent information, for as long as reasonably necessary (which may be longer than the period described in the table above) to:
- enforce or investigate potential violations of our Terms of Use, this Privacy Policy, or any Foundation or user community-based policies;
- investigate and defend ourselves against legal threats or actions;
- help protect against vandalism and abuse, fight harassment of other users, and generally try to minimize disruptive behavior on the Wikimedia Sites;
- prevent imminent and serious bodily harm or death to a person, or to protect our organization, employees, contractors, users, or the public; or
- detect, prevent, or otherwise assess and address potential spam, malware, fraud, abuse, unlawful activity, and security or technical concerns.

Audits and improvements

The Foundation is committed to continuous evaluation and improvement of these guidelines, and to periodic audits in order to identify such improvements. As we make changes to existing and systems, we will update these guidelines to reflect our changing practices.

Design of new systems

In order to support these data retention periods and our overall privacy policy, new tools and systems implemented by the Foundation will be designed with privacy in mind. This will include:

inclusion of these data retention guidelines as requirements during the design process;
legal consultation during the design and development process; and
inclusion of privacy considerations in the code review process.

Ongoing handling of new information

Despite our best efforts in designing and deploying new systems, we may occasionally record personal information in a way that does not comply with these guidelines. When we discover such an oversight, we will promptly comply with the guidelines by deleting, aggregating, or de-identifying the information as appropriate.

Contact us

If you think that these guidelines have potentially been breached, or if you have questions or comments about compliance with the guidelines, please contact us at privacywikimedia.org.

[1] For the purposes of this table, "user account" means username, user ID, or IP address; "reader" means visitor to a Wikimedia project.

[T 1]