In mid-February, OpenAi unveiled Sora, its generative AI model that creates videos from text instructions and still images.
In an interview with the Wall Street Journal, OpenAi Chief Technology Officer Mira Murati confirmed that the new AI will be available later this year.
In recent days, the Italian privacy guarantor has started an investigation to understand how OpenAi trains the Sora algorithm, what user data it collects and processes and whether it uses external sources to generate the videos.
Murati reported that the software will be able to create not only movies but also the sound part, at least in a subsequent update of the model, making the result something very close to a cinematographic work.
To safeguard the transparency of content, videos should also contain metadata and labels capable of making users understand when what they are watching is real or the result of AI.
"Considering the possibility of creating misinformation" continues the manager "we are committed to taking all the right preventive measures before a public release of Sora, also in view of the presidential elections in November".
According to Murati, Sora is much more expensive to run than other AI models, but OpenAi aims to make it available at "similar costs" to Dall-e, the AI that creates static images from text, or the famous chatbot ChatGpt.
"These are tools designed to expand man's creative possibilities. The roles will change but in the end the global economy will benefit from them" the words of the CTO.
Reproduction reserved © Copyright ANSA