An AI-generated advert by Toys “R” Us has been panned for including a host of errors.

The TV ad, the first to be created using OpenAI’s text-to-video tool Sora, debuted at the Cannes Lions Festival this week.

The toy firm said its production company had been granted special access to Sora, which is not yet publicly available.

The advert begins by depicting a young Charles Lazarus, the late founder of Toys “R” Us, in his family’s bicycle shop before moving into a dream sequence.

“Charles Lazarus was a visionary ahead of his time and we wanted to honour his legacy with a spot using the most cutting-edge technology available.” said Kim Miller Olko, Toys “R” Us global chief marketing officer.

Rife with errors

At first glance, the technology appears to generate life-like characters in an authentic setting but, on closer inspection, the video is rife with mistakes.

The most glaring of which surrounds the character of young Charles, who is almost immediately seen moving his head from one side to another in an unnatural way.

His spectacles also appear to have bamboozled the AI tool. The glasses are initially seen with a bridge around two-thirds of the way up but, by the end of the video, a second bridge has appeared connecting the top of each lens.

Similarly, the freckles on Charles’ face which feature in the video have disappeared by the final shot of him.

Background details are also problematic, with bicycles in the family shop merging into one another in an unrealistic manner.

In the dream sequence, toys on the middle shelf appear clearer than those on the shelf above despite being located the same distance from the camera.