Written evidence submitted by Tumblr (SMH0171)

About Tumblr

Tumblr is a blogging and publishing platform for amateur and professional authors and creators.  In mid-2017, Tumblr became part of Oath, a house of media and technology brands.

Tumblr Community Guidelines and report handling

Tumblr’s Community Guidelines clearly set out standards of behaviour and content that users are required to adhere to.  These are deliberately clear and in colloquial language to be accessible to audiences of all ages.  These guidelines are under constant review and were most recently updated last month.

Content which breaches Tumblr’s Community Guidelines or Terms of Service comes to the attention of Tumblr in a variety of ways, including:

●        Tumblr provides several ways for users to report violations of the Community Guidelines, including via links in the Community Guidelines, Help Pages and FAQs, as well as the ability to directly flag the content within the product.  Tumblr has an active dialogue with users about community issues – for example, via the Tumblr Staff blog – and this dialogue encourages reporting by users so that moderators are able to take appropriate action consistent with the Community Guidelines. 

●        Tumblr uses technical tools to flag certain types of content for review.  For example, Tumblr scans user uploads using IWF and NCMEC hashes of known CSAM and partners with Google to deploy CSAI Match to detect known CSAM video. 

●        Tumblr is part of the industry hash-sharing consortium which deploys a database of known terrorist imagery (videos and images) that can be used by members to identify and review content that violates company policies. 

●        Tumblr accepts and prioritises reports from expert partners, including the Revenge Porn Helpline and the Internet Watch Foundation.

Tumblr has a team of highly-trained Trust and Safety professionals who review reports of content that may violate Tumblr policies.  The recent revision of Tumblr’s Community Guidelines equips these moderators with clear guidelines to inform the review and removal of hate speech and content that glorifies violence. The update also eliminates any ambiguity in Tumblr’s zero-tolerance policy on non-consensual sexual content. An overwhelming amount of care and nuance is needed to evaluate reports fairly and accurately.  Tumblr has recently increased the size of its content review team to ensure that it can continue to apply this level of scrutiny to incoming reports.  Tumblr engineers, legal staff and community engagement teams are also involved in these processes.  Tumblr prioritises reports about content or conduct that are most harmful to users and responds accordingly. 

Managing inappropriate content

In designing and developing the product, Tumblr has sought to balance the essence of a publishing service – as you note in your letter, it is not a wholly logged-in experience like social networking – with an acknowledgement that some content permitted on Tumblr can be inappropriate for certain audiences.  To address this, Tumblr has developed Safe Mode, which filters sensitive content in a user’s Dashboard and search results, with specific settings for minors.  Safe Mode is turned on by default for all Tumblr users. Users that are over 18 years old may choose to turn Safe Mode off, however, users who are under 18 years of age cannot turn Safe Mode off.  Users must be logged in to view the Tumblr dashboard and to use the Tumblr application. Content can only be viewed by logged out users on the Tumblr blog network and logged out users cannot see blogs that are marked as "explicit" on the blog network

Tumblr has a team of dedicated engineers and also employs partnerships with third parties to develop its tools and capabilities.   Tumblr is continuing to invest heavily in technology that will detect and block content that violates Tumblr’s Community Guidelines at the time of upload, before it is publicly viewable.  While this technology shows great promise, it cannot replace human review. Fortunately, Tumblr benefits from an engaged and passionate user base that proactively flags content that appears to violate Tumblr’s Community Guidelines.  These reports are carefully reviewed by Tumblr’s expert content moderation staff. 

In addition, Tumblr displays interstitial “PSAs” (public service announcements) to users in response to searches containing certain trigger words such as those related to suicide, depression, eating disorders, self-harm and domestic violence.  These PSAs interrupt the user experience and provide links to expert sources of advice.

We take seriously the issues raised in your letter and the opportunity to explain our practices and procedures.  We are committed to fostering a safe, secure and respectful environment on Tumblr and we have endeavoured to reply to your questions as fully as possible.  We hope these answers are helpful to your Committee’s inquiry.

September 2018