OpenAI has officially launched its highly-anticipated video-generating tool, Sora, but with a notable caveat: a key feature that allows users to generate videos using uploaded photos or footage of real people as references will not be available to most users, at least not yet. The company has opted to take a cautious approach to rolling out this feature, citing concerns over potential misuse and the need for further testing.

The decision to hold back this feature is likely a response to the growing concerns over deepfakes and misinformation. According to data from ID verification service Sumsub, deepfake fraud worldwide has increased by more than 10 times from 2022 to 2023. OpenAI acknowledges the potential for abuse, stating that it is taking an "incremental approach" to learn from early patterns of use and fine-tune its approach to safety.

Despite the limitations, Sora still offers a range of features that make it a powerful creative tool. Users can generate videos from text prompts or images, edit existing videos using a Re-mix tool, and create sequences of videos using a Storyboard interface. A Blend tool also allows users to combine two videos into a new one that preserves elements of both. Subscribers to OpenAI's ChatGPT Pro and Plus plans will have access to Sora, but only if they live in certain countries.

In an effort to prevent misuse, OpenAI has implemented several measures, including a filter to detect whether a generated video depicts someone under the age of 18. If it does, the company applies a "stricter threshold" for moderation related to sexual, violent, or self-harm content. Additionally, all Sora-generated videos will contain metadata that shows their provenance, adhering to the C2PA technical standard. This metadata can be removed, but OpenAI is pitching it as a way for platforms that support C2PA to quickly detect whether a video originated from Sora.

OpenAI is also taking steps to address copyright concerns, which have been a major point of contention for AI companies. The company is using "prompt re-writing" to prevent Sora from generating videos in the style of a living creator. This move is likely a response to the numerous lawsuits filed by artists against AI companies, including OpenAI, alleging that they have trained on their works without permission to create AI tools that regurgitate content in their unique styles.

As the AI-generated video landscape continues to evolve, OpenAI's cautious approach to rolling out Sora's full features is a nod to the complexities and challenges involved. While the tool has the potential to be a powerful creative force, it also raises important questions about safety, ethics, and responsibility. As the company continues to fine-tune its approach, it will be interesting to see how Sora is received by the creative community and how it will be used in the future.