The social networks are failing to detect hate speech
Everyone is entitled to their opinion. But when free speech goes to extremes and uses damaging vocabulary, it becomes an extremely dangerous tool. As social media use grows, so does online hate. Groups and individuals have become targets because of their ethnicity, religion, gender, and sexual orientation.
Unfortunately, many social platforms lack the solutions to moderate hate speech before it causes irreparable harm.
The link between social networking sites and hate
Social media is a convenient way to reach millions of users across the world. However, the moderation guidelines are still limited in their effectiveness, despite the growing use of toxicity to target minority groups. For example, when there is no moderation, violent marketing campaigns can go viral in moments and affect many people.
Anyone can post on a social platform anonymously and within seconds the post can go viral. Manually monitoring bullying and aggression on social networks is almost impossible, both in terms of the time it takes and the possibility of human error. Not to mention the effect that reading hundreds of racist or homophobic messages can have on the moderator.
The big platforms use machine learning algorithms to detect hate speech.
Currently, social platforms like Facebook detect only 20-40% of hateful content (in English only), which means up to 60% is still there targeting vulnerable people.
The impact of hate being posted on social media
There are many examples of the destructive impact hateful content can have on communities. This can range from inciting others to commit violence, promoting propaganda, separating groups to emotional abuse, and even encouraging murder and suicide.
One example is the ‘Fetrah’ campaign on Twitter. This anti-LGBT (lesbian, gay, bisexual, and transgender) movement spread rapidly, despite Facebook and Instagram banning the account. The logo is a blue and pink flag that represents the male and female genders and rejects any other communities. The campaign was created by three Egyptian marketers and seeks to harm gay and transgender people and groups, particularly in the Middle East.
A final example was Facebook’s inability to detect hate in the East African language Kiswahili (Swahili). A series of ads were submitted to Facebook that referred to rape, beheadings, and extreme physical violence by the non-profit group Global Witness. The ads were approved by Facebook (some were even in English). While the ads were never published, as this was a test, this shows a gap in content moderation on social media.
These examples demonstrate that giant social platforms are still not monitoring and filtering out hateful content. Every user deserves maximum protection through the use of reliable content moderation systems.
Whilst 40% of users leave a platform after their first exposure to toxic language, many stay and participate, resulting in collective abuse. Research also shows that even if a user reports hateful comments, one-third are not deleted. This leaves hate speech targets vulnerable and shows the need for improved moderation standards on social media platforms that have a major influence on users’ psyches.
How can Bodyguard help you and your communities?
We believe in freedom of speech and the right to do a job without receiving harmful comments.
Bodyguard.ai has developed a unique moderation model that is both instantaneous and intelligent and which combines the speed of intervention provided by a machine with the subtle ability of a human to assess and evaluate.
The Bodyguard solution:
detects and moderates toxic content in real-time
protects individuals, communities, and staff members
prevents negative exposure (bad buzz…)
is tailored to suit the customer’s needs
As a more effective method than employing a human moderator, Bodyguard’s objective is to have a positive social impact on our society. Toxic content is detected in real-time and is moderated immediately, eliminating the possibility of human error.