- A groundbreaking new tool named Sora, described as ‘terrifying,’ has been introduced by Open AI, capable of generating hyper-realistic videos based on text input, sparking concerns among experts.
- Experts caution that this advanced artificial intelligence tool could potentially disrupt entire industries like film production and contribute to the proliferation of deep fake videos.
- This development coincides with the upcoming crucial US presidential election and numerous other elections worldwide in 2024.
By Alice Wright For Dailymail.Com
Open AI, the organization responsible for ChatGPT, has revealed a new tool named Sora, which has been described as ‘terrifying’ due to its ability to create hyper-realistic videos based on text prompts, leading to warnings from experts.
Sora, unveiled by Open AI on Thursday, showcases impressive capabilities such as drone footage capturing Tokyo in a snowy setting, waves crashing against the cliffs of Big Sur, and a grandmother enjoying a birthday celebration.
Experts caution that this new AI tool could potentially disrupt entire industries such as film production and contribute to the proliferation of deep fake videos, particularly concerning in the lead-up to the pivotal US presidential election.
‘Generative AI tools are advancing at a rapid pace, compounded by the prevalence of social networks, creating a vulnerability in our democracy at a critical juncture,’ remarked Oren Etzioni, the founder of TruMedia.org, in an interview with CBS.
‘At a time when we are grappling with these challenges, we are facing one of the most significant elections in history,’ he added.
The quality of AI-generated images, audio, and video has shown significant advancements over the past year, with companies like OpenAI, Google, Meta, and Stable Diffusion racing to develop more sophisticated and user-friendly tools.
OpenAI explains on its website, ‘Sora can construct intricate scenes featuring multiple characters, specific movements, and detailed elements of both the subject and the background.’
‘The model not only comprehends the user’s prompt but also interprets how these elements manifest in the physical world.’
The tool is presently undergoing testing and evaluation to assess potential safety risks, with no confirmed release date for public use at this time.
While the examples provided by the company are currently non-offensive, experts express concerns that this technology could usher in a new era of highly realistic deepfake content.
‘We are essentially building this aircraft while it’s in flight, and it’s scheduled to land in November, if not sooner, without the Federal Aviation Administration, without historical precedence, and lacking the necessary tools,’ cautioned Etzioni.
Sora ‘will facilitate easier creation of high-quality video deepfakes by malicious actors, offering them enhanced flexibility to produce videos that could be exploited for malicious purposes,’ stated Dr. Andrew Newell, the chief scientific officer of iProov, a firm specializing in identity verification.
‘Voice actors or individuals creating short videos for video games, educational purposes, or advertisements will likely be the first to feel the impact,’ Newell warned.
The proliferation of deepfake videos, including those of a sexual nature, poses an escalating challenge for both private individuals and public figures.
Sora represents a significant technological milestone!
However, the real game-changer lies in AI-generated virtual worlds
Envision playing as an ant or any evolving character with a captivating narrative
Incredibly optimistic about the future of AI tech 🌊🌊🌊🤖 pic.twitter.com/vEvJWa0Tdn
— Kibshi — AI Meme Coin — meme/acc (@KiboShib) February 18, 2024
#SoraAI is pushing boundaries to a whole new level 😨
Prompt: “We are filming a promotional video for a political campaign. Showcasing an Indian minister, a symbol of transformation, traversing the town and catalyzing development. The clip should vividly portray the dramatic progress and change he brings to the city.” pic.twitter.com/hH20AK66Gy
— Agha (@Drfunfish) February 17, 2024
‘Consider the progress we’ve witnessed in just a year of image generation. Where will we be in another year?’ pondered Michael Gracey, a film director and expert in visual effects, in an interview with the Washington Post.
‘We will be implementing several crucial safety measures before integrating Sora into OpenAI’s products,’ the company assured.
‘We are collaborating with red team experts—specialists in areas such as misinformation, offensive content, and bias—who will rigorously test the model from an adversarial perspective.’
Additionally, the company stated, ‘We are developing tools to identify misleading content, including a detection classifier capable of discerning videos generated by Sora.’
Earlier this year, deepfake images garnered significant attention when AI-generated explicit images of Taylor Swift circulated on various social media platforms.
These images, originating from the website Celeb Jihad, depicted Swift engaged in explicit acts while adorned in Kansas City Chiefs attire within a stadium setting.
The singer expressed anger and contemplated legal action in response to the unauthorized dissemination of these manipulated images.
President Joe Biden has also addressed the issue of AI manipulation and disclosed instances where he fell victim to deepfake impersonations using his voice.
‘This is already occurring. AI technologies are being utilized for deceptive purposes. Deepfakes leverage AI-generated audio and video to tarnish reputations,’ Biden remarked, highlighting the potential for spreading misinformation and perpetrating fraud.
‘With AI, fraudsters can fabricate a three-second audio clip of your voice. I’ve seen one of myself a couple of times – I thought, ‘When did I say that?“’ shared Biden during a public address.
He further emphasized the technology’s capability to deceive individuals through fraudulent schemes. IT experts have also underscored the risks associated with the misuse of AI in political contexts.
On Friday, several prominent technology companies committed to taking ‘reasonable precautions’ to prevent the misuse of artificial intelligence tools in disrupting democratic elections globally.
Executives from Adobe, Amazon, Google, IBM, Meta, Microsoft, OpenAI, and TikTok pledged to implement preventive measures.
‘Everyone acknowledges that no single tech company, government, or civil society organization can address the challenges posed by this technology and its potential malicious applications independently,’ remarked Nick Clegg, Meta’s president of global affairs, following the agreement signing.
Meanwhile, Nvidia, a leading provider of computer chips utilized in AI technologies, has experienced a significant surge in its market value. Last week, Nvidia’s stock reached \(781.28 per share at one point, resulting in a market capitalization of \)1.78 trillion.
This valuation surpassed Amazon’s market cap of $1.75 trillion.
According to CNBC, this marks the first instance since 2002 where Nvidia’s market worth exceeded that of Amazon post-market closure.
The California-based company has observed a remarkable 246% increase in its stock value over the past year, driven by the growing demand for its AI server chips, which can command prices exceeding $20,000 per unit.
