Tagged: cyberhate

Facebook and Periscope introduce new strategies to fight hate speech

by Martin Schmalzried

It’s happening. COFACE has been asking for a long time for new strategies to fight hate speech and cyberbullying, including the use of algorithms and Artificial Intelligence to assist human moderators and also the recourse to community based moderation where users can vote on taking content down.

Facebook has been steadily developing/training algorithms to help identify offensive photos. At this point, algorithms report more potentially offensive photos than humans!  Twitter, which has been attacked in the past over its lackluster ability to fight abuse, has also invested in Artificial Intelligence to help weed out abusive content. The help and assistance of AI is welcome, especially since relying on human moderation only comes with many sets of problems such as speed of reaction or even negative psychological consequences for the human moderators which are forced to look at the worst content humanity has to offer.

Progress in the development of such algorithms could benefit all online service providers as Facebook has vowed to share its findings more broadly [1]

On the other hand, Periscope (owned by Twitter) is rolling out another approach to moderation, adapted to live streaming and instant feedback and user interaction: a form of live community based moderation.  Viewers will be able to immediately report a comment they deem abusive during a live streaming.  The app then randomly selects within the audience a “jury” who will vote whether the comment is abusive/inappropriate or not.  Should the “jury” vote to sensor the comment, the author of the comment will not be able to post temporarily.  If the author of the comment repeats the offense, he/she will be blocked from commenting for the rest of the live streaming [2]

Although such initiatives are long overdue, COFACE welcomes their introduction and will closely follow their development, hopefully contributing to create a better online environment for children and for families.

Advertisements

Building partnerships in tackling Cyberhate

The International Network Against CyberHate (INACH) recently held its annual conference which brought together key players from civil society, law enforcement and the industry to discuss how Cyberhate could best be tackled through partnerships. The presentations brought to light several interesting points on how to tackle cyberhate.

From law enforcement, the presentations showed that the very process of fighting cyberhate is a very time consuming and lengthy one. Websites that spread hate can take several months to several years to be taken down, especially for international websites. Removing such content from the internet often requires the clearing of many legal hurdles such as a clear mandate from a court, an official request from one law enforcement agency to another in case of international websites etc. This basically means that for individuals seeking for redress or that have been directly targeted by a specific website, the process can be extremely long and complex.

From the industry represented by Facebook and Twitter, most of the solutions involve reporting followed up by moderation and take-down. More recently, many social networks including Facebook and Twitter have started partnering up with local civil society organisations and providing them privileged access to reporting tools and the moderation team to take down material faster. Another recent trend is the encouragement of counter-speech, namely individuals that retaliate against negative messages with a flood of positive ones.

Finally, the civil society organisations presented many activities and good practices from their respective networks, ranging from tools to be used in classrooms to coordination and cooperation work on an international level.

During the conference, we intervened to underline the specifics of cyberbullying. One of the most important problems in tackling cyberbullying online is the timing of an intervention. By the time a reporting has been filed and a moderation team has processed it, much of the damage and impact on the victim has already been done and new offensive or hurtful material has been published.

The industry should take moderation a step further and involve users as volunteer moderators directly in order to speed up the moderation and review process. Many online services have already adopted such models each tailored to the services’ needs: Wikipedia and its voluntary contributors/editors or the “tribunal” in the online game League of Legends.

For more information about IN@CH, visit their website