Instagram’s recommendation algorithm can rapidly funnel ordinary users from mainstream self-improvement content toward antisemitic conspiracy theories and extremist propaganda within days, according to a new report by a U.S.-based antisemitism watchdog.
“You don’t have to search for antisemitic content to find it on Instagram,” said Oliver Marks, research director at the Antisemitism Research Center (ARC), the research arm of the Combat Antisemitism Movement, which released the study. “Our findings show that users engaging with normal self-improvement posts are algorithmically guided toward virulent antisemitic narratives and conspiracy theories. When platforms optimize for engagement without sufficient safeguards, they can end up amplifying hate to vast audiences.”
Researchers created two Instagram accounts designed to simulate ordinary users interested in self-improvement content. The accounts interacted only with mainstream, non-political posts and were monitored during three consecutive days of 45-minute browsing sessions.
One account focused on wellness and biohacking content, while the other followed fitness and discipline creators. Researchers recorded all recommended videos during each session.
The wellness-focused account was shown 59 classifiable videos, more than 32% of which were categorized as coded or explicit antisemitic content, including nine such videos during a single session on the third day.
The fitness-focused account received 71 classifiable videos, with 24% identified as antisemitic content, including 17 in one browsing session.
Among the recommended videos were conspiracy theories claiming that the Israel-Iran war was intended to “distract from a globalist plot to make Americans allergic to meat.” Other content promoted unfounded theories about off-grid living, alleged digital ID systems as tools of government control, and false claims about food products and historical events.
Additional videos featured rants about “Jewish financial power” and claims that the Rothschild family orchestrated the sinking of the Titanic.
Researchers noted that antisemitic content appeared during the first browsing session, before any meaningful engagement history had developed, suggesting early-stage algorithmic exposure.
The report identified three key trends: speed, volume and convergence. Speed referred to the near-immediate appearance of antisemitic content. Volume described the growing proportion of such material over time. Convergence referred to different user profiles being directed toward similar antisemitic narratives and conspiracy theories despite starting with different interests.
The report follows earlier ARC investigations into online antisemitism, including the identification of more than 80 AI-generated fake “rabbi” accounts on Instagram that promoted antisemitic content. Similar activity was also reported on YouTube and TikTok, with many accounts later removed by Meta following reports from the Combat Antisemitism Movement.
The organization said the findings underscore the need for greater transparency and accountability in social media recommendation systems and called on Meta, Instagram’s parent company, to review its algorithms and implement stronger safeguards.
