😯 Elon Musk's controversial Kamala Harris Deepfake

Also: Runway AI trained on thousands of YouTube videos

Author

Evolving AI
July 29, 2024

Welcome, AI enthusiasts

A manipulated video mimicking Vice President Kamala Harris's voice to falsely attribute statements to her is raising concerns about the potential of AI to mislead, with election day just three months away. According to a report, Runway trained its AI text-to-video generator on thousands of YouTube videos and pirated films. Additionally, OpenAI could face losses of up to $5 billion this year due to the high costs of running AI products like ChatGPT. Let’s dive in!

In today’s insights:

  • Elon Musk's controversial deepfake Kamala Harris video

  • Runway trained on thousands of scraped YouTube videos  

  • OpenAI to lose $5B this year alone

Read time: 4 minutes

🗞️ LATEST DEVELOPMENTS

Evolving AI: Elon Musk's post of an AI-generated Kamala Harris video sparks debate over synthetic media ethics and regulations.

Key Points:

  • Elon Musk shared an AI-manipulated video of Kamala Harris on X without a disclaimer.

  • The video violates X’s guidelines on synthetic media that could cause public confusion.

  • The creator labeled it as a parody, but Musk's post lacked this context, sparking criticism.

Details:

Elon Musk posted a video on X featuring an AI-generated voice imitating U.S. Vice President Kamala Harris, making false statements without indicating it was fake or a parody. The video claims Harris made controversial remarks about President Biden and herself, causing confusion among viewers. The original creator labeled it as an "ad parody," but Musk's post omitted this disclaimer, leading to significant backlash. The video seemingly violates X's guidelines, which require synthetic media to be flagged if it could cause public confusion or harm. Despite this, neither Musk nor the platform has addressed the issue, and the post remains uncorrected.

Why It Matters:

While some participants in X's "Community Notes" feature have suggested labeling Musk's post, no such label had been added as of Monday morning. Some users questioned whether his post might violate X's policies, which state users "may not share synthetic, manipulated, or out-of-context media that may deceive or confuse people and lead to harm." By not making viewers aware that the content they’re watching is AI-faked, this may lead to serious issues in a time when the U.S. is already politically divided.

Is AI going to have an impact on the 2024 US elections?

Login or Subscribe to participate in polls.

Source: 404 Media

Evolving AI: A leak reveals that Google-backed AI video generator, Runway, trained on stolen content.

Key Points:

  • Runway's Gen-3 Alpha trained on pirated films and YouTube content.

  • Internal spreadsheet shows use of popular content from Disney, Netflix, and Sony.

  • AI companies facing increased scrutiny for copyright violations.

Details:

Runway's Gen-3 Alpha video generator has come under fire for using pirated content and ripped YouTube videos in its training data, according to a leaked internal spreadsheet. The spreadsheet lists videos from major YouTube channels, including those owned by Disney, Netflix, and Sony, as well as links to websites hosting pirated content. Despite claims of being trained on videos and images, the exact sources were not disclosed by Runway. A former employee revealed that proxies were used to avoid being blocked by YouTube during data scraping. Runway, which raised $141 million last year from investors including Google, is not alone; other AI companies like OpenAI have also faced accusations of using unlicensed content for training their models.

Why It Matters:

We are in the Wild West of training and sourcing data for these models. AI startups are likely to see many lawsuits coming their way, as questions arise about whether they can just take any data to train their models. However, they might be in a strong position to win these lawsuits. Earlier this month, a California judge dismissed nearly all claims in a lawsuit accusing GitHub, Microsoft, and OpenAI of copying code from developers.

Source: Techspot

Evolving AI: OpenAI could lose as much as $5 billion this year due to the massive costs of running AI products like ChatGPT — and likely needs to raise more money within the next 12 months.

Key Points:

  • OpenAI's projected 2024 spending includes $7 billion on AI training and $1.5 billion on staffing.

  • Despite $2 billion in annual revenue, high costs may necessitate another funding round.

  • Confident in doubling revenue by 2025, OpenAI continues to expand with projects like SearchGPT.

Details:

According to The Information, OpenAI could face up to $5 billion in losses next year, risking bankruptcy within 12 months. The company, led by Sam Altman, is predicted to spend about $7 billion on AI training and another $1.5 billion on staffing in 2024. These figures surpass those of competitors like Anthropic, which anticipates a $2.7 billion burn rate. Although OpenAI generates around $2 billion annually and is valued at over $80 billion, the significant expenses might require additional funding within a year. Despite this, OpenAI is expanding, recently launching its AI-powered search engine, SearchGPT, currently in limited release.

Why It Matters:

Multiple news outlets are reporting that OpenAI might be on the brink of bankruptcy in 12 months because of their losses—but that’s hard to believe. OpenAI has become such an important player in the world of tech and AI that they’ll have no trouble raising funds when the time comes for it. It does raise serious questions about how deep the losses can go before investors will shy away from investing in AI when it's clearly not profitable and probably won’t be for a long time.

💡 Tip of the Day

Two of the most well-known and important figures in tech and AI will join forces for an intriguing keynote today.

Jensen Huang and Mark Zuckerberg will discuss the future of AI and simulation and the pivotal role of research. The two leaders will explore how fundamental research is enabling AI breakthroughs and how generative AI and open source will empower developers and creators. Be sure to tune in!

🎯SNAPSHOTS

Direct links to relevant AI articles.

🚦 Google: Google is using AI in an attempt to improve traffic flow.

🤖 LLMs: Llama 3.1 vs GPT-4o vs Claude 3.5: A Comprehensive Comparison of Leading AI Models.

⚡️ Generative AI: Generative AI requires massive amounts of power and water, and the aging U.S. grid can’t handle the load.

📈Trending AI Tools

  • 💹 Profit Leap - A tool for business intelligence and strategic insights (link)

  • 🎥 Depthify.ai - A tool to convert videos into 3D spatial videos (link)

  • 🎞️ Potion - A tool to create personalized videos at scale for sales prospecting (link)

  • 💡 Inline Help - A tool for contextual support and knowledge directly within a website or app (link)

  • 🔄 PipeLime - A tool to automate sales funnels with lead generation and deal closures (link)

Reply

or to participate.