SAN FRANCISCO — OpenAI, an artificial intelligence company, unveiled its latest innovation: “Sora,” a cutting-edge AI tool capable of generating lifelike 60-second videos based on a simple text prompt. This advancement marks a significant leap in AI video technology and deepfake creation, raising concerns about the potential for deceptive practices.
Initially, “Sora” will have limited access, catering to a select group of artists, filmmakers, and researchers known as “red teamers,” who evaluate the tool’s susceptibility to misuse, OpenAI revealed in a recent announcement.
Building upon the technology powering OpenAI’s image-generating tool, DALL-E, “Sora” takes a user’s prompt, refines it into detailed instructions, and leverages an AI model trained on video and images to produce the final video output.
The quality of AI-generated content, spanning images, audio, and video, has seen rapid advancements in recent times. Companies such as OpenAI, Google, Meta, and Stable Diffusion are in a race to enhance the capabilities of these tools and explore commercial opportunities. However, concerns have been raised by democracy advocates and AI experts regarding the potential misuse of such technology to deceive the public.
For instance, in Pakistan, AI was utilized to create a virtual version of former prime minister Imran Khan delivering speeches while he was incarcerated. Similarly, an advertisement supporting Florida governor Ron DeSantis utilized AI-generated audio mimicking the voice of former president Donald Trump during his presidential campaign.
Despite efforts by tech companies to regulate the use of these tools, challenges persist in enforcing such policies effectively. OpenAI, for instance, took action against a developer who created a bot resembling Democratic candidate Dean Phillips only after media scrutiny. This incident underscores the ongoing struggle to monitor and control the misuse of AI-generated content, particularly in the political realm.
The rapid evolution of this technology has sparked interest across various industries, prompting stakeholders in Hollywood and the news sector to assess its potential impact on their respective fields.
AI-generated clips from “Sora” showcase diverse scenarios, from a grandmother blowing out birthday candles to dogs playing in the snow, each brought to life through AI interpretation of textual prompts. Notably, the realism achieved in these videos, particularly in depicting everyday scenes, surpasses the current capabilities of many other AI platforms.
