AI Trace
Recommendation SystemConsumer FacingVerified

Reviewed and published by trentmaziarz, March 24, 2026. Discovered and drafted by our automated research pipeline.

The Instagram Reels tab is powered by AI that continuously selects and ranks short videos for each user, mostly from accounts they do not follow. The system strongly favors videos people watch all the way through and share with others via direct message. Instagram published a detailed explanation in June 2023.

Details

The Reels recommendation pipeline gathers candidate videos, uses a lightweight machine learning model to narrow them to approximately 100, applies integrity filters to remove potentially policy-violating content, and then applies a ranking model to order the remaining videos. Watch time is the most heavily weighted signal — especially watching a video more than once or through to completion. Direct message shares are counted as three to five times more valuable than a like when determining how broadly a video is recommended. A September 2025 investigation by the Molly Rose Foundation found the algorithm still directed self-harm content toward test accounts posing as 15-year-old girls.

Products affected

Instagram Reels

Sources & Evidence

Cite this record

Trace Foundation. (2026). Instagram: The Instagram Reels tab is powered by AI that continuously selects and ranks short videos for each user, mostly from accounts they do not follow. The system strongly favors videos people watch all the way through and share with others via direct message. Instagram published a detailed explanation in June 2023 (data as of 2026-03-24) [Data set record]. AI Trace. https://www.aitrace.org/r/practice/a833112e-075d-4336-9f3c-c748dcbd6d8b. Accessed October 13, 2026.

Stable link
https://www.aitrace.org/r/practice/a833112e-075d-4336-9f3c-c748dcbd6d8b
Data as of
March 24, 2026
Last verified
March 24, 2026

How to cite AI Trace

Other practices by Instagram

Have evidence about Instagram's AI practices? Submit a report.

Report a Sighting →