Detect Deepfakesby Resemble AI
Deepfake case study · Multi-modal

TikTok AI pulled after Charli D’Amelio 'blueberry' video - - Happy Mag

TikTok scaled back its AI "overviews" feature after it incorrectly described videos, including one of Charli D’Amelio as "a collection of various blueberries."

Incident date
May 2026
Target
Charli D’Amelio
Updated May 7, 2026 · 1 min read

In May 2026, TikTok pulled back its new artificial intelligence "overviews" feature after the tool began misinterpreting videos in ways that led users to question its accuracy. The experimental feature, intended to summarize videos, suggest related content, and explain on-screen activity, instead generated inaccurate descriptions.

What happened

One video of Charli D’Amelio was described as "a collection of various blueberries with different toppings." A dog training video was described as "a captivating display of intricate origami art." A Shakira promotional video was summarized as "a repetitive sequence of several distinct blue shapes appearing and moving across the screen." TikTok stated the tool is being refocused to identify products in videos. The feature had been available to limited users in select regions and was powered by a mix of TikTok’s own systems and third-party AI models.

Sources