Sora characters.
Photo credit: OpenAI

OpenAI is introducing “character cameos” in Sora, a new feature allowing users to create reusable characters from their own videos. The update, announced on October 29, 2025, also adds video “stitching” and new “leaderboards.”

The new creative tools arrive just over a week after a NewsGuard analysis found that Sora 2 produced realistic videos advancing provably false claims 80 per cent of the time when prompted.

The new character cameo feature allows users to create a character by uploading a video or selecting an existing Sora video, which can then be tagged and reused. Users can set permissions to keep characters private, share them with mutual followers, or make them public.

“More sophisticated narratives”

The stitching feature, available on the app and website, allows users to select and connect multiple clips in their drafts. “Stitching allows you to create more sophisticated narratives and extend your story, with granular control over each segment,” OpenAI stated in the release.

The NewsGuard analysis, published on 22 October 2025, tested Sora using 20 false claims, generating videos for 16 of them. The report also found that Sora’s watermark can be easily removed using free online tools.

In response to the NewsGuard findings, OpenAI spokesperson Niko Felix said the videos “appear to violate OpenAI’s usage policies, which prohibit misleading others through impersonation, scams, or fraud”. He added that “we take action when we detect misuse”.

Leave a Reply

Your email address will not be published. Required fields are marked *

You May Also Like

Journalism students ‘get out of the bubble’ to rebuild public trust

Journalism is facing dual challenges of lost trust and declining relevance. But…

Chatbots are helping human users ‘hallucinate’ their own reality

While much attention has focused on AI “hallucinating” false facts, a study…

Cities are at a breaking point. Here is how ‘Physical AI’ can fix them

With two-thirds of the world’s population soon to live in urban areas,…

To govern AI, we must stop policing software and start capping ‘compute’

Trying to regulate subjective AI capabilities is a losing battle. Instead, we…

Why the AI job apocalypse might just be history repeating itself

From silent film stars to bank tellers, professions threatened by new technology…