TEL AVIV, ISRAEL | CyberWell has alerted major social-media platforms to an antisemitic trend in which imagery drawn from Japanese folklore is being used to demonize Jews and evade content-moderation systems.
The Tel Aviv-based nonprofit said the trend uses images of Tengu, supernatural beings traditionally depicted with red faces, long noses and small box-like hats.
According to CyberWell, users are placing the figures alongside photographs of visibly Orthodox Jews and established antisemitic caricatures to promote stereotypes about Jewish appearance and portray Jews as demonic, dangerous or evil.
The organization said the content has circulated online since at least 2023 but has gained renewed visibility and engagement in recent weeks across Facebook, Instagram, TikTok and X.
Posts have appeared in English, French and Farsi.
CyberWell said one Instagram post received more than 40,000 likes, while an X post amplifying the comparison accumulated more than 249,000 views.
Antisemitism Without the Word โJewโ
Most of the posts identified by CyberWell reportedly do not contain the words โJewโ or โJewish.โ
Instead, users rely on visual comparisons, recognizable stereotypes and images already associated with antisemitic culture to communicate the message.
The technique may allow the posts to evade automated moderation systems that depend heavily on written language, keywords and explicit slurs.
CyberWell said some posts connect Tengu imagery with the โHappy Merchant,โ a widely circulated antisemitic caricature portraying Jews as greedy, manipulative and conspiratorial.
Other posts reportedly use the figures to portray Jews as warmongers or accuse them of kidnapping children, echoing longstanding antisemitic myths.
โComparing Jews to evil demons through this use of Japanese imagery is a grotesque form of dehumanization, with roots in religious antisemitism,โ said Tal-Or Cohen Montemayor, CyberWellโs founder and CEO.
โRepeat exposure to the dehumanization of Jews through powerful algorithmically driven platforms has seeded the conditions in which hostility, hatred, discrimination and even violence against Jews are increasingly justified and commonplace today.โ
She warned that the repetition and algorithmic amplification of dehumanizing imagery can desensitize audiences to increasingly explicit expressions of anti-Jewish hatred.
โThe Holocaust did not begin with mass murder, but with years of propaganda, discrimination and systematic state-sponsored dehumanization of Jews,โ Cohen Montemayor said.
โThe increasingly open anti-Jewish vitriol that CyberWell has tracked online has similarly desensitized many of our most open societies to antisemitism.โ
Cultural References Repackaged as Hate
The Tengu trend follows another coded antisemitic narrative identified by CyberWell involving the term โCelestial Dragon.โ
The term originates in the Japanese manga and anime series One Piece. Antisemites have appropriated it to promote conspiracy theories about Jewish wealth, political influence, and global control.
Neither Tengu imagery nor the original One Piece reference is inherently antisemitic. The problem arises from the context in which users deliberately connect those cultural references to Jews and established anti-Jewish stereotypes.
CyberWell said the two trends demonstrate how familiar cultural material can be stripped of its original meaning and repackaged to carry old antisemitic narratives in new, less immediately recognizable forms.
The approach is sometimes described as โalgospeak,โ a method of using substitute words, images, symbols or coded expressions to avoid automated moderation.
โEspecially with the widespread adoption of AI and the rise of algo-speak, antisemites do not need to use the word โJewโ to create antisemitic content,โ Cohen Montemayor said.
โMuch of the hate content is designed to evade content moderation systems. Technology must adapt.โ
Platforms Urged to Recognize Visual Codes
CyberWell said the identified posts violate platform rules prohibiting dehumanizing comparisons, harmful stereotypes and claims that members of protected groups are inherently dangerous.
Metaโs hateful-conduct policy defines prohibited content as direct attacks against people based on protected characteristics, including religious identity. Its rules include certain forms of dehumanizing speech and harmful stereotyping.
However, enforcement becomes more difficult when a post communicates its message through visual association rather than explicit language.
โAs platforms increasingly rely on AI-based enforcement, their systems must recognize the images and visual cues that make coded content antisemitic,โ Cohen Montemayor said.
โExpert partners who understand how antisemitism evolves across languages, cultures and contexts can help platforms identify emerging forms of Jew-hatred before they spread.โ
CyberWell has called on social-media companies to improve their ability to detect visually coded antisemitism, recognize recurring stereotypes and apply existing hate-speech policies more consistently.
The organization also wants platforms to keep working with specialists who can identify how antisemitic narratives evolve across cultures and languages.
Hate Adapts to Technology
The warning follows broader CyberWell research into the growth of AI-generated antisemitic content.
In a separate report, the organization analyzed 307 AI-generated antisemitic posts that collectively attracted more than 30 million views and 2.8 million interactions.
The research found that coded language, humour, false news formats, synthetic imagery and disclaimers such as โsatireโ were used to disguise hateful messages and complicate moderation.
TikTok removed 88.2 percent of the content in that dataset, according to CyberWell, while Meta removed 67 percent. The reported removal rates were considerably lower for YouTube and X.
CyberWell cautioned that removing material eventually does not prevent harm when posts remain online long enough to reach large audiences.
Its online research argues that generative AI is making it easier to create persuasive antisemitic images, videos and audio at scale while allowing old conspiracy theories to be adapted into new formats.
The Tengu trend illustrates the same challenge without requiring users to state their intended target openly.
The folklore changes. The platform changes. The code changes.
The antisemitic message remains the same.
TheJ.ca receives no government money. Independent journalism depends on readers who believe these stories must be told. Consider supporting our efforts.
Stand Proud. Stand Strong. Stand Zionist.
