Deepfakes Of Taylor Swift Spread All Over The Internet + Microsoft Forms New Team For Next-Gen, Cost-Effective AI + Netflix's Concerns Over Generative AI In Entertainment Production
Good morning!
Welcome to the 61st edition of the Quantumics Weekly Roundup.
In this edition, we’ll take a deep dive into the latest AI trends.
And as usual, we’ll explore the latest news and information in data and AI, business, and tech.
Let’s go!
Exploring The Real-world Impact of AI-Driven Image-Based Exploitation
Millions of people recently encountered manipulated and fake explicit AI-generated images featuring Taylor Swift on social media, highlighting the urgent need for regulations on potential malicious uses of AI technology. The White House Press Secretary expressed concern over the incident, stating that they are disturbed by the dissemination of fabricated images. Emphasizing the role of social media companies, she mentioned that while these companies have autonomy in content management, they should actively enforce their own rules to curb the spread of misinformation and non-consensual, intimate imagery involving real individuals.
Jean-Pierre emphasized recent actions taken by the administration, such as launching a task force to combat online harassment and abuse, along with the Department of Justice introducing the first national 24/7 helpline for survivors of image-based sexual abuse. Despite these efforts, there is currently no federal law in the U.S. specifically addressing the creation and dissemination of non-consensual deepfake images.
Surprising and upsetting fans, the absence of such legislation became evident. However, Rep. Joe Morelle has recently renewed efforts to pass a bill criminalizing the nonconsensual sharing of digitally altered explicit images at the federal level. The proposed legislation includes penalties such as imprisonment and fines.
Expressing hope that the Taylor Swift incident will generate momentum and support for their bill, a spokesperson for Morelle stated that the legislation, known as the "Preventing Deepfakes of Intimate Images Act," is bipartisan and currently under consideration in the House Committee on the Judiciary.
The term "image-based sexual abuse" encompasses not only the sharing of non-fabricated intimate images but also deepfake pornography. In the past, generating AI-generated content required a certain level of technical expertise, but with recent advancements in AI technology, it has become as simple as downloading an app or clicking a few buttons.
Experts now point to the existence of a thriving commercial industry that specializes in creating and disseminating digitally manipulated content, giving the appearance of sexual abuse. Some websites dedicated to these fakes have amassed thousands of paying members.
Notably, a town in Spain gained international attention when young schoolgirls reported receiving manipulated nude images of themselves. These images were produced using easily accessible "undressing apps" powered by artificial intelligence, sparking a broader conversation about the potential harm inflicted by such tools.
Microsoft's Quest for Cost-Efficient AI Development
Microsoft is reportedly in the process of creating artificial intelligence technology that is more compact and cost-effective than the existing technology it currently acquires from OpenAI. As per The Information's report, Microsoft has established a new team dedicated to developing conversational AI that demands less processing power than what OpenAI offers. The report also highlights the transfer of several key developers from Microsoft's research group to this newly formed Gen AI team.
Expected to be integrated into the Azure cloud unit, the team may be overseen by Microsoft corporate vice president Misha Bilenko. Historically, Microsoft heavily relied on OpenAI for its AI-based product development. For instance, Microsoft's Bing Chat utilizes the GPT-4 language model, and the Bing AI Image Creator employs the Dall-E 3 image generator for crafting AI-generated images.In the realm of artificial intelligence, Microsoft recently introduced the Copilot Pro subscription, offering users access to AI-powered Office features. This subscription provides priority access to the latest language models, including OpenAI's GPT-4 Turbo. Additionally, users gain the capability to create a personalized Copilot version called Copilot GPT, and there are improvements in image quality through the Bing AI image generator.
In a related development, Microsoft has unveiled CoPilot for Microsoft 365, catering to enterprise customers and aiming to empower businesses with AI capabilities. The company has also announced the release of the Copilot Android and iOS app, expanding the accessibility of these AI features.
Netflix's Annual Report Highlights Generative AI Risks and Uncertainties
Netflix further expressed concerns about the use or adoption of new and emerging technologies, stating that it "may increase our exposure to intellectual property claims, and the availability of copyright and other intellectual property protection for AI-generated material is uncertain."
While generative AI-created entertainment is still in its early stages, there is an acknowledgment that the landscape can evolve rapidly. Notably, widely accessible AI tools are capable of suggesting storylines, character arcs, and dialogue, with some systems, like ChatGPT, even capable of generating a basic script based on prompts.
Generative AI played a role in the creation of the 2022 film 'Everything Everywhere All at Once,' and reports suggest the existence of generative AI systems that can produce videos. AI is also being utilized for data-driven predictions regarding how unconventional storylines may impact viewer reactions.
In May of the previous year, the Writers Guild of America (WGA) initiated a strike over streaming services' impact on writers' income. One of their demands was for production companies to "regulate the use of material produced using artificial intelligence or similar technologies."