Reports describe antisemitic material appearing in Instagram account tests, including one Nexus Project sample in which 14 of 15 recommended videos were described as antisemitic. Separate accounts of searches and content moderation measured different outcomes, so their findings belong to their own tests.
Keep recommendation results separate from removal counts: one describes what an account was shown, the other what Instagram did after users reported content.
What Instagram account tests reported
Futurism reported that searches for “synagogue,” “Jew,” “Jews” and “Erika” on a decoy account returned antisemitic material. When the account was checked the next day, its feed contained more hateful content, Futurism said.
The Nexus Project’s public summary described a separate account that followed only Instagram’s official account. After its researcher watched suggested videos, nearly the entire feed became antisemitic in less than an hour.
What the 15-video sample counted
Religion News Service reported a specific Nexus Project recommendation sample: within an hour of searching “synagogue” on a new account, a researcher received 15 recommended videos. Fourteen were antisemitic, and nine came from verified accounts. Those counts describe the recommended videos in that test; the Project’s broader feed description was a separate observation.
What the ADL test measured
The Anti-Defamation League (ADL) tested Instagram’s handling of user reports, not its recommendations. In January and February 2026, ADL researchers submitted 150 accounts and 103 posts. Instagram removed 11 of the accounts and eight of the posts; in 20 cases, it said it lacked the capacity to review the reports, according to ADL’s account of the test.
The recommendation tests describe material surfaced in account feeds and searches. The ADL figures describe outcomes after reports were submitted. Each measure answers a different question.
What Meta said
Meta disputed the Nexus Project’s characterization of its standards and said its commitment to tackling antisemitism and its related protections were unchanged. Meta also said the prevalence of hateful-conduct violations on its platforms had remained consistent since 2025, Futurism reported.