Tech CEO Shows Shocking Deepfake Of Kari Lake At Hearing On AI Impact On Elections

Forbes Breaking News
18 Apr 202408:40

TLDRRidel Gupta, a tech entrepreneur and founder of Deep Media, addresses a hearing on the impact of AI on elections, emphasizing the growing threat of deep fakes. He explains that deep fakes are AI-manipulated images, audio, or video that can mislead or harm, and outlines the importance of understanding generative AI technologies such as Transformers, GANs, and diffusion models. Gupta highlights the rapid improvement and decreasing cost of deep fake production, which could lead to a significant portion of online content being fake by 2030. He discusses the political implications, including the use of deep fakes for political assassination or to make politicians appear more relatable. Gupta stresses the need for a collaborative effort among various stakeholders to combat the issue. His company, Deep Media, is actively involved in developing solutions, working with journalists, researchers, and platforms to detect and mitigate the spread of deep fakes. He concludes by demonstrating how their technology can identify deep fakes, setting a gold standard for detection in the tech community.

Takeaways

  • 👥 The speaker, Riddhiman Gupta, is a tech entrepreneur focused on combating deep fakes with his company Deep Media.
  • 🧠 Deep fakes are synthetically manipulated AI images, audio, or video that can mislead or harm, and do not include text.
  • 🤖 Key technologies behind generative AI include the Transformer, Generative Adversarial Network (GAN), and diffusion models.
  • 💰 The cost to produce deep fake videos is decreasing rapidly, potentially reaching 1 cent per minute by 2030.
  • 📈 Deep fakes are improving in quality and are expected to make up a significant percentage of online content, nearing 90% by 2030.
  • 🔍 Deep fakes have already impacted elections, with manipulated videos of political figures being used for political assassination or to sway public opinion.
  • 🚨 The real threat of deep fakes lies in their potential to undermine trust in real content, leading to plausible deniability and misinformation.
  • 🌐 A collaborative effort involving government, generative AI companies, platforms, journalists, and deep fake detection companies is needed to address the issue.
  • 🛡️ Deep Media is part of various initiatives and collaborations aimed at detecting and combating deep fakes, including the DARPA Semafor and AI Force program.
  • 📊 The company's platform can deliver solutions at scale across image, audio, and video, with a focus on minimizing false positives and negatives.
  • 📈 The speaker emphasizes the importance of staying ahead in the cat-and-mouse game between deep fake creation and detection technologies.
  • 📚 The presentation includes a demonstration of how AI analyzes and detects deep fakes, highlighting the complexity and evolving nature of the technology.

Q & A

  • What is the name of the person who founded Deep Media?

    -The person who founded Deep Media is Riddhiman Gupta.

  • What is a deep fake?

    -A deep fake is a synthetically manipulated image, audio, or video created using AI that can be used to harm or mislead.

  • What are the three fundamental technologies that generative AI is based on?

    -The three fundamental technologies that generative AI is based on are the Transformer, the Generative Adversarial Network (GAN), and the Diffusion Model.

  • Why is it important to define deep fakes?

    -Defining deep fakes is important because it helps to understand the type of synthetic media content that has the potential to dismantle society by misleading or harming people.

  • What is the current cost of producing a deep fake video per minute?

    -The current cost of producing a deep fake video is about 10 cents per minute, which is rapidly decreasing towards 1 cent.

  • What is the potential percentage of content on online platforms that could be deep fakes by 2030?

    -The percentage of content on online platforms that could be deep fakes is approaching as much as 90% by 2030.

  • How do deep fakes impact political elections?

    -Deep fakes impact political elections by creating false narratives, such as fake announcements or endorsements, which can lead to political assassination or manipulate public opinion.

  • What is the biggest threat posed by deep fakes according to Riddhiman Gupta?

    -The biggest threat posed by deep fakes is not the fake content itself, but the doubt it casts on real content, leading to a situation where people can no longer trust the authenticity of any media they consume.

  • What are the five groups of people that need to work together to solve the deep fake problem?

    -The five groups of people that need to work together to solve the deep fake problem are government stakeholders, generative AI companies, platforms, investigative journalists, and deep fake detection companies.

  • What is the role of Deep Media in addressing the deep fake issue?

    -Deep Media is committed to solving the deep fake problem by developing technology to detect deep fakes, collaborating with news organizations, participating in initiatives like the DARPA Semafor and AI Force program, and working with big tech platforms to adopt detection technology.

  • How does Deep Media approach the detection of deep fakes?

    -Deep Media approaches the detection of deep fakes by using advanced AI technology to analyze and learn from the patterns in both real and fake media, ensuring a low rate of false positives and false negatives.

  • What is the significance of the deep fake of Kari Lake shown during the hearing?

    -The deep fake of Kari Lake is significant because it demonstrates the high quality of current deep fake technology, using proprietary generative models that are not publicly released, and it was correctly identified as a deep fake by Deep Media's detection system.

Outlines

00:00

💡 Introduction to Deep Fakes and Their Impact

Ridel Gupta, the founder of Deep Media, introduces himself as a tech-savvy entrepreneur and hacker who started developing apps and websites at a young age. He discusses his academic background in machine learning from Yale and his subsequent focus on generative AI, which led to the founding of Deep Media in response to the emerging threat of deep fakes. Gupta emphasizes the importance of defining deep fakes as AI-manipulated media that can mislead or harm, and he outlines the need for understanding three key technologies behind generative AI: Transformers, Generative Adversarial Networks (GANs), and diffusion models. He also highlights the rapid improvement and decreasing cost of deep fake production, the potential societal impact, and the political manipulation that deep fakes can enable. Gupta stresses the need for a collaborative effort among various stakeholders to address the deep fake problem.

05:01

🛠️ Solutions to the Deep Fake Problem

Gupta, as a believer in the free market and the potential for AI to be a force for good, views deep fakes as a market failure and a societal issue that requires legislative action. He presents a vision where proper regulation can help mitigate the negative impact of deep fakes. Gupta shares insights into how AI perceives media, which is crucial for effective detection. He provides a demonstration of his platform's capabilities, showing its ability to differentiate between real and fake content with high accuracy, minimizing false positives and negatives. The presentation includes examples of how AI analyzes audio and video, and how Deep Media's technology can detect even high-quality deep fakes. Gupta concludes by emphasizing the dual role of his company as both a creator and a detector of deep fakes, setting a high standard in the ongoing battle against this form of misinformation.

Mindmap

Keywords

💡Deepfake

A deepfake refers to a synthetically manipulated image, audio, or video created using AI that can be used to mislead or harm. In the context of the video, deepfakes are a significant concern because they have the potential to disrupt society by creating mistrust in media and affecting political processes. An example from the script is the mention of deepfakes involving political figures like President Biden, Trump, and Hillary Clinton.

💡Generative AI

Generative AI is a branch of artificial intelligence that involves creating new content, such as images, audio, or video, that did not exist before. It is the technology behind deepfakes. The speaker emphasizes the importance of understanding generative AI to address the deepfake problem, as it is rapidly improving and becoming more accessible.

💡Transformer

A Transformer is a type of AI architecture that is fundamental to generative AI. It is used in the creation of deepfakes and is one of the three key technologies mentioned by the speaker that underpins the majority of generative AI models. The script does not provide a detailed explanation of how Transformers work but emphasizes their importance in the context of deepfakes.

💡Generative Adversarial Network (GAN)

A GAN is a type of machine learning model consisting of two parts, a generator and a discriminator, that work together to produce and improve synthetic data. It is one of the core technologies behind deepfakes. The speaker mentions GANs as a fundamental technology in the creation of deepfakes, highlighting their role in the development of synthetic media.

💡Diffusion Model

A diffusion model is another key technology in generative AI, used to generate high-quality synthetic data. It is one of the three fundamental technologies mentioned by the speaker. The script does not go into the technical details of diffusion models but includes them as a critical component in the creation and detection of deepfakes.

💡False Positive and False Negative Rate

In the context of deepfake detection, a false positive occurs when real content is incorrectly identified as a deepfake, while a false negative is when a deepfake is missed and classified as real. The speaker emphasizes the importance of maintaining a very low false positive and false negative rate for effective deepfake detection systems.

💡Political Assassination

The term 'political assassination' in the script refers to the use of deepfakes to damage the reputation or credibility of political figures. The speaker mentions examples of deepfakes involving President Trump and Hillary Clinton, which were created to mislead the public and potentially influence political outcomes.

💡Plausible Deniability

Plausible deniability is a situation where someone can claim they are not responsible for an action because they cannot be proven to be involved. In the context of the video, the speaker warns that deepfakes can lead to a scenario where politicians or others can falsely claim that real content is a deepfake, thus undermining trust in genuine information.

💡Silicon Valley

Silicon Valley is a region in California known for its high-tech innovation and development. The speaker mentions Silicon Valley in the context of adopting a solutions-oriented approach to the deepfake problem, indicating that technology companies and innovators there are actively working on solutions to combat deepfakes.

💡Content Authority Initiative

The Content Authority Initiative is a collaborative effort involving companies like Adobe, aimed at labeling real and fake content to help users distinguish between them. The speaker mentions their participation in this initiative as part of the broader effort to combat the spread of deepfakes and misinformation.

💡DARPA Semaphor and AI Force Program

The DARPA Semaphor and AI Force Program is a research initiative that brings together researchers, corporations, and government resources to address the deepfake problem. The speaker's company is part of this program, which is focused on developing technologies and strategies to detect and mitigate the impact of deepfakes.

💡Externality

An externality is an economic term referring to the cost or benefit that affects a party who did not choose to incur that cost or benefit. In the video, the speaker describes misinformation and fraud caused by deepfakes as a negative externality, suggesting that proper legislation could help internalize this negative impact and foster a healthier AI ecosystem.

Highlights

Ridel Gupta, a tech entrepreneur and hacker, founded Deep Media to combat the rise of deep fakes.

Deep fakes are synthetically manipulated AI images, audio, or video with potential to harm or mislead.

The human mind is particularly susceptible to manipulation by image, audio, and video content.

Three fundamental technologies behind generative AI are Transformer, Generative Adversarial Network (GAN), and diffusion model.

Deep fakes are becoming high-quality, cheap to produce, and increasingly prevalent on online platforms.

The cost of video deep fakes is rapidly decreasing, nearing 1 cent per minute.

Deep fakes have already impacted elections, with manipulated videos of political figures causing significant harm.

The larger threat of deep fakes lies in their potential to undermine trust in real content.

Deep Media is collaborating with government, media, and AI companies to develop solutions to the deep fake problem.

Deep Media has assisted journalists from CNN, Washington Post, and Forbes in detecting and reporting on deep fakes.

The company is part of the DARPA Semaphor and AI Force program aimed at addressing the deep fake challenge.

Deep Media is also involved in the Content Authority initiative, which focuses on labeling real and fake content.

Gupta emphasizes the importance of a free market approach and the potential for AI to be used for good.

Deep fakes represent a market failure and a tragedy of the commons, which can be mitigated through proper legislation.

Deep Media's platform can deliver solutions at scale across image, audio, and video to combat deep fakes.

The company's technology can accurately detect deep fakes without misidentifying real content as fake.

An example of a high-quality deep fake featuring Kari Lake was presented, demonstrating the capabilities of Deep Media's detection system.

Deep Media uses its own generative AI models to stay ahead in the cat-and-mouse game against deep fakes.

Gupta calls for a collaborative effort among various stakeholders to set a gold standard for deep fake detection.