A report by the Nexus Project says new Instagram accounts were quickly shown antisemitic posts and recommendations, including AI-generated content. Futurism says its own tests found similar material; Meta disputes the report’s account and says its protections against antisemitism remain in place.
Watch Desk analysis
What happened
The Nexus Project says it created new Instagram accounts and searched for content related to Jewish people. In one test, the first result for “synagogue” was an AI-generated video, and within less than an hour the researchers’ homepage was largely antisemitic content. The report also says Instagram recommended antisemitic conspiracy topics beneath posts.
Futurism reports that it repeated searches using a decoy account and found hateful material among the results, including AI-generated videos. It also reports that Meta said the Nexus Project’s account contains factual inaccuracies and misrepresents its standards. Meta says it prohibits hateful attacks based on religion and that its commitment to tackling antisemitism is unchanged.
Why it matters
The report describes a troubling combination: AI-generated material can be produced quickly, while recommendations may steer people towards more of the same. If that pattern holds, the issue is not only what users upload, but what the platform’s discovery systems surface to people who have just arrived.
The evidence here is attributed reporting, not a complete independent audit of Instagram. Futurism’s reported tests add a separate account of what searches returned, while Meta’s response disputes the report’s conclusions. Those distinctions matter, especially when claims concern both platform safety and the treatment of a targeted community.
Our read
This deserves more than a passing mention because it puts a concrete question to a platform’s recommendation systems: what does a new account get shown, and how quickly? The reported results are alarming; the next useful evidence is a clear account of the tests, the platform’s response, and whether changes to recommendations or enforcement follow.
What to watch
- Whether the Nexus Project publishes its methods and the full results of its account tests.
- Whether Meta provides further detail or evidence supporting its response.
- Whether subsequent testing finds the same patterns across different searches and new accounts.
Discussion spark: When a platform disputes findings that its recommendations are amplifying hate, what evidence should it publish to show that its systems are working?
Sources and evidence
- Source update (16 November 2026, 17:01 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.