GIFCT’s Hash-Sharing Database enables GIFCT member companies to quickly identify, and share signals, of terrorist and violent extremist activity in a secure, efficient and privacy-protecting manner.
How does hash-sharing work?
Known as perceptual hashes, a hash is a numerical representation of original content (video, image, PDF) that cannot be reverse-engineered to recreate the content. These hashes are then added to the database with a series of labels to help other members understand what content corresponds to the hash, including content type, terrorist entity that produced the content, and its behavioural elements. A GIFCT member can then select a hash to see if it identifies and matches to visually similar content on their platform. As a result, members are sharing signals about terrorist content they have identified on their platform so that other members can quickly identify if the same content is shared on their platform and assess it in line with their policies and terms of service. All without sharing any user data between companies.
When a member reviews content corresponding to a hash they can use the feedback tool in the database to tell us whether they agree or disagree that the hash is correctly labeled and meets GIFCT’s taxonomy, or criteria for inclusion. At GIFCT, we respect that each member might operate a little differently. We don’t tell our members how to use the hashes or how to apply their own policies. Rather, we are here to help our members collaborate, and together we can make terrorists ineffective online.
Hash-Sharing Database Taxonomy
Hashes added to the database must fit within GIFCT’s Taxonomy, otherwise understood as inclusion criteria. This taxonomy addresses content based on a terrorist or violent extremist entity producing the content and the type, or behavioral elements, of the content the entity produced.
This taxonomy contains a series of parameters for inclusion in the database that takes into account:
- the producers of the content being terrorist and violent extremist entities
- the types of terrorist or violent extremist behavior associated with the content and/or the offline violence the content depicts and/or relates to
- categorize the hashes shared in terms of the form of the content in terms of images, videos, PDFs, and URLs
The current taxonomy of the database reflects several evolutions and expansions over time. The taxonomy originally addressed images and videos produced by entities on the United Nations Security Council’s Consolidated Sanctions List and the behavior of the content that entity had produced.
In 2019, the first expansion was made to include content produced by the perpetrators of an offline attack live streaming or recording their violence, related to GIFCT’s Incident Response Framework.
Most recently, in response to GIFCT’s 2021 Taxonomy Report comprising recommendations from global experts and our tech company members, the taxonomy has expanded to include terrorist and violent extremist attacker manifestos and URLs that direct people to where the content addressed in GIFCT’s taxonomy is hosted.
Expanding the taxonomy iteratively means GIFCT can empower our members to combat a wider range of terrorist activity, work to address the Islamist extremist-bias that currently exists in the larger counterterrorism field, and remain diligent to impacts on the human rights of those most vulnerable in this context.