- February 16, 2024
- AI Projects
As recently as Thursday, the creators of the ChatGPT phenomenon launched a revolutionary AI feature that instantly creates short videos from text prompts.
San Francisco-based tech firm OpenAI has unveiled its new text-to-video app called Sora. This innovation is not the only one on the market; Tech giants like Google, Meta, and startup Runway ML have also demonstrated technology with similar capabilities.
The accuracy of the videos produced by OpenAI — some of which were made after CEO Sam Altman’s social media request for creative text ideas — left viewers surprised while raising concerns about potential moral and social implications.
One creative pitch submitted at X by a freelance photographer from New Hampshire was: “Show me an elderly social media influencer doing a homemade gnocchi workshop in a vintage kitchen in Tuscany with atmospheric cinematography.” Altman soon showed a convincingly realistic video that matched the description.
To date, this tool has not been made publicly available, and OpenAI has only provided a brief overview of its development process. OpenAI faced a lawsuit from various authors and The New York Times over the alleged use of copyrighted texts in the ChatGPT training. They have not disclosed the source of the images and video footage used for Sora’s training, although they are compensating the Associated Press an undisclosed amount for the use of their written news content.
In a blog post, OpenAI mentioned its intention to consult with artists, politicians, and other stakeholders before making Sora available to everyone. “The firm is working with ‘red teams’ — disinformation specialists, hate speech and bias — who will thoroughly review the system,” OpenAI said. “In addition, the company is developing tools to identify deceptive content, such as classifiers that can determine whether Sora created a certain video.”
