New Israeli study exposes coded language used by anti-semites on social media

March 2, 2021 by TPS
Read on for article

Anti-Semites are employing the use of a secret language, or “dog whistles,” to evade detection by algorithms on social media tracking anti-Semitism, posing challenges to law enforcement and social media platforms, a University of Haifa study has revealed.

Candles shaped as the Star of David at the ceremony marking the Holocaust Martyrs’ and Heroes’ Remembrance Day (Yom Hashoa), at Hadassah Medical Center in Jerusalem. Jerusalem, May 2, 2019. Photo by Avigail Eyal/TPS

The new study, authored by Dr Gabriel Weimann, a professor of communication at the University of Haifa and web intelligence analyst Ari Ben-Am, has revealed the existence of a new coded language on social media being used by anti-Semitic groups to fly under the radar of artificial intelligence-designed algorithms.

This sophisticated language, which is used for both propaganda and recruitment purposes, is hidden in plain sight and encourages violent incidents like the Christchurch shooting in New Zealand or the January Capitol Hill riots in Washington D.C.

The study explores the emergence of this new language, its characteristics, transmission, and usage, and its findings are intended to serve both law enforcement and private sector hi-tech organizations in their quest to combat hate speech online.

Given the popularity of the platform, the study focused primarily on images and messages disseminated on Facebook.

Posts used for the analysis were identified qualitatively after scanning hundreds of relevant groups and pages, eventually reaching a smaller sample of tens of pages with a minimum of 500 followers or likes as an initial seed group.

Much of this coded language, the study found, was shown to rely on the use of “dog-whistles,” a coded message communicated through words or phrases commonly understood by a particular group of people, but not by others.

The method could be as simple as swapping one word for another such as far-right users calling Jews “Skypes,” African-Americans as “Googles” and Latinos as “Yahoos.”

Some tactics to disguise the anti-Semitic communications use numerology where numbers stand for people or concepts. The number 88, for example, stands for “Heil Hitler,” since H is the eighth letter in the alphabet.

Finally, the language also employed visual cues often by manipulating already popular memes found in pop culture. This method is by far the most sophisticated one as an algorithm’s ability to detect hate speech within an image is still limited.

Weimann is working on a comprehensive glossary of terms chronicling instances of this coded language and what it means, and hopes this study will encourage law enforcement and hi-tech companies to be more vigilant and aware that their algorithms are not the be-all and end-all of ending hate speech.

“It’s clear that security, counter-terrorism, and government agencies, as well as social media platforms, are doing much to crack down on abuse,” Weimann said.

However, “we need to educate the operators of these companies that run social media platforms to report these violations and also teach their users how to spot them. A human eye is still much savvier than a computer-generated algorithm,” he said.

So successful is this new language based on numbers, acronyms, and hidden images, that it has driven the formation of alternative social media sites such as Gab, Bitchute, and imageboards such as 4Chan, 8Chan, and Neinchan.

“While this is purely an academic study, it has real-world implications,” Weimann added. “Being Internet savvy and understanding what you’re seeing online is important. We need to learn what they do, slow them down, and reduce their activity and efficiency. Words kill even if in a new language.”

TPS

Speak Your Mind

Comments received without a full name will not be considered
Email addresses are NEVER published! All comments are moderated. J-Wire will publish considered comments by people who provide a real name and email address. Comments that are abusive, rude, defamatory or which contain offensive language will not be published

Got something to say about this?

This site uses Akismet to reduce spam. Learn how your comment data is processed.