Social media applications. Photo: Public Domain.

A new technology startup is offering social media companies an AI (artificial intelligence) powered tool for becoming more effective at removing antisemitism and Holocaust denial from their platforms at a time when anti-Jewish hatred is spreading rapidly online.

Only 23 percent of antisemitic content and 36 percent of content denying the Holocaust is deleted from Twitter, Facebook, Instagram, and others, CyberWell founder and CEO Tal-Or Cohen told The Algemeiner on Wednesday. The total is even less for content posted in Arabic, falling to ten percent.

The problem, she said, is that English speaking users are likely to report antisemitic posts. In Arabic speaking countries and others in which antisemitism is normalized, users are not and the content is seldom detected or deleted. Platforms in Germany and France are also teeming with antisemitic content, a report by the Institute for Strategic Dialogue (ISD) said in 2021. Telegram, a company started by two Russian entrepreneurs in 2013, contained two-thirds of all antisemitic keywords found during the group’s data analysis.

“Platforms will only devote resources to keeping users safe if enough people report problems,” Cohen explained. “Social media platforms’ over relying on user reports and under-investment in combating Jew-hatred online in non-English languages has allowed blatant hate to proliferate in Arabic. By alerting them to these data insights, we hope to direct their attention to fixing the problem.”

Founded in 2022, CyberWell markets itself as the “world’s first live database of online antisemitism.” Using the International Holocaust Remembrance Alliance definition of antisemitism, its automated tool peruses the world wide web to find social media posts that, for example, use the hashtag “Holohoax.” Searching for that term in English yields no results and directs users to resources about the Holocaust, but searching it in Arabic prompts pages of the worst antisemitic conspiracies, including that Israel “invented” or “exaggerated” the Holocaust. Cohen hopes Cyberwell can change that.

The startup is entering a $65 billion market fraught with complications. AI bots, programmed to mimic human thought processes, have at times been permeated by antisemitic tropes.

In Aug. 2022, BlenderBot 3, a chatbot created by Meta Platforms Inc. — the parent company of Facebook and Instagram — told a Wall Street Journal reporter that conspiracies alleging that Jews control the economy are “not implausible,” and in 2016, Microsoft’s chatbot, called “Tay,” was taken offline for praising Adolf Hitler and generating other racist and misogynistic comments. Experts have said that the technology is still in its infancy. Teaching it not to emulate the worst antisocial human tendencies is a long term project. Cyberwell’s AI, designed to draw from the most widely used definition of antisemitism, represents a positive step towards harnessing the power of AI technology to reflect what’s best in humanity.

To date, Cyberwell has found over 3,000 antisemitic posts from 86 countries and in six languages. Cohen noted that CyberWell wants to collaborate with social media companies not denounce them. Its automated tool, which keeps a database of the content it identifies, can report to them what it finds, supplementing their efforts to maintain adherence to community standards and hate speech policies. The work, she noted, is important as Jewish and non Jewish organizations around the world observe Holocaust Remembrance Day on Friday and reflect on rising antisemitism across the world.

“The social media platforms are consistently exposing people on a very large scale to antisemitism, and now we are seeing the results” she added. “The conversation about fighting antisemitism absolutely has to be centered on the issue of digital policy compliance and social media accountability.”