Instagram’s recommendation algorithm can easily direct users from mainstream self-improvement content to virulent antisemitic material and Nazi propaganda, according to an investigation published Wednesday by an antisemitism watchdog.

For the project, the Antisemitism Research Center of the Combat Antisemitism Movement (CAM) used two neutral persona accounts following mainstream creators in the field of fitness and wellness, engaging in three daily sessions of 45 minutes online.

To prevent algorithmic bias, researchers strictly observed the feeds without liking, sharing or commenting on any posts, the report said.

Without any active user intent, both accounts were rapidly served content promoting conspiracy theories and explicit hate speech.

By the third day, 31 percent of the wellness account’s content and 18% of the fitness account’s content consisted of explicit antisemitism, CAM said.

Over the three days, 32% of wellness videos and 24% of fitness videos contained coded or explicit antisemitism.

“Despite different starting points, both accounts were independently routed toward the same antisemitic narrative clusters, tropes, scapegoating logic, and, in several cases, the same specific pieces of content,” the report noted. “This suggests the issue is not only a community-level phenomenon but a deeper failure in the algorithmic architecture.”

The report is intended as an “exploratory, hypothesis-generating piece of research,” not a statistically valid survey, with several methodological limitations,........

© The Times of Israel