- TikTok has been testing AI summaries for its videos
- Feature returns wildly inaccurate text subtitles
- TikTok says it will now retire the technology
In the age of AI deepfakes, it’s a good idea to treat everything you see on social media with some degree of skepticism, but TikTok’s misinformation problem has been made worse by some wildly inaccurate AI captions, and it’s bad enough that the video platform is now scaling back this captioning technology.
As Business Insider reported, TikTok had been testing AI-based text summaries for videos with a limited number of users. However, after numerous errors and hallucinations, the technology will be limited to identifying products in videos, rather than fully describing the content of the video.
Those errors and hallucinations included describing a video of celebrity Charli D’Amelio speaking to the camera as showing a “collection of various blueberries with different toppings” and labeling a dog training video as “a captivating display of intricate origami art, meticulously folded from a single sheet.”
You don’t have to look far on social media for more examples: There’s what appears to be an image of two cats with the caption “a person showing off an impressive new robotic arm with multiple dexterous fingers,” for example.
‘Trash that has nothing to do with the video’
I’m so glad Tiktok added this new AI overview feature. I’m not sure how I would survive the app without it. pic.twitter.com/Y5P31nridiMay 3, 2026
It’s unclear exactly what has gone wrong and is causing TikTok’s AI summaries to get the wrong idea so often (although presumably the feature worked at least part of the time). Recognizing the content of images and videos is usually something that AI can do quite reliably.
However, that clearly hasn’t been the experience of many TikTok users. One Redditor described the subtitles as “completely out of place,” while another said they were watching “garbage that has nothing to do with the video,” and the AI summary also served to distract from the actual title of the video.
Other online examples show a video of a Kentucky Derby horse race described as “showing an intricate piece of calligraphy” and a cooking video with an overhead shot of a gray frying pan labeled “a single ball bouncing and rolling on a green surface,” although these screenshots could also be fake, of course.
Even as AI is introduced into more and more applications and devices, hallucinations and errors remain a major problem, one that AI companies do not like to admit. Whether it’s a TikTok video or a legal document, if you want the AI to summarize something, it would be wise to perform additional checks.
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds.
