OpenAI Will Not Allow Sora to Generate Celebrities’ Faces and Voices

Metaverse Planet October 21, 2025Last Updated: October 21, 2025

1 minute read

OpenAI has decided to prevent Sora from generating videos containing actors’ voices and likenesses. This decision comes after growing concerns and the creation of deepfakes without consent.

OpenAI officially introduced its video generation model, Sora 2, in recent weeks. The new model, which comes with audio support, quickly went viral on social media but also led to the emergence of some controversial content.

The number of deepfake images featuring celebrities had increased significantly. Now, Sora has received a significant update regarding this. OpenAI has decided to increase Sora’s security measures. The AI model will no longer generate videos containing the voices and likenesses of celebrities.

The Decision Came After Growing Concerns from Actors

The US-based actors’ union SAG-AFTRA, along with Bryan Cranston (known from Breaking Bad) and several other organizations, released a joint statement. The statement mentioned that they collaborated to ensure such a decision was made.

Many actors, including Cranston, had expressed concerns about the increase in deepfake images of celebrities on the Sora 2 and Sora applications. It was underlined that this was being done without the consent of the actors. We had seen a similar situation for historical figures such as Martin Luther King Jr. Following the reactions from King Jr.’s family, OpenAI blocked the creation of such content.

With the new decision, only the likenesses of people who have given permission will be able to be created with Sora. We know that artificial intelligence causes great concern in this regard. Unfortunately, as video models advance, the number of such contents increases, and artists’ reactions intensify. If the necessary precautions are not taken, we may see deepfakes reach uncontrollable levels.