Google's Jigsaw open sources its anti-harassment tool to help protect journalists

It can help users easily detect, document and hide toxic comments on social media.


Anyone can be harassed online, but it's an inevitability for activists and journalists covering contentious topics and challenging their country's authorities — especially if they're women. Google's Jigsaw unit has released the open source code for a web app called Harassment Manager that can help them review, sort and export the toxic comments they get on social media. Developers can build on the code to create their own version of the tool, but Jigsaw does have implementation partners, including Thomson Reuters Foundation that's rolling out a functional web app for its journalists in June.

In its current form, Harassment Manager works on Twitter, which teamed up with Jigsaw for the project. It uses the unit's Perspective API to detect comments that are most likely to be harmful and allows users to easily document and hide those responses, as well as to mute or block the accounts harassing them. Jigsaw has been training Perspective to detect online harassment since at least 2017 by feeding it comments identified as toxic by human reviewers.

As The Verge reports, Perspective gives Harassment Manager the power to check the messages a user gets and to sort them into high, possible and unknown levels of harmfulness. The user can then process them in batches instead of dealing with them one by one, and they can blur toxic messages if they don't want to read them.

According to Jigsaw's research, 70 percent of female journalists receive threats online, and more than 40 percent of them stopped reporting their story as a result. The unit is hoping that the code can give developers what they need to be able to create a tool to protect people who "deal with disproportionately high toxicity online," especially "female journalists, activists, politicians and other public figures." Jigsaw says open sourcing the tool is just a step towards helping people who constantly deal with online harassment. It will continue working with NGOs in the journalism and human rights space to figure out how Harassment Manager's technology can help keep their communities safe.