A star OpenAI engineer says he’s starting a new team to chase artificial superintelligence

**OpenAI Star Engineer Will DePue Returns to Lead New Team Aiming for Artificial Superintelligence**

Will DePue, one of the core contributors behind OpenAI’s text-to-video model, Sora, announced on X this Wednesday that he is returning to OpenAI after a long break to start a new team. The newly formed group also includes fellow Sora engineers Troy Luhman and Eric Luhman. Their mission is an “incredibly high-risk bet” with a small but significant chance of achieving artificial superintelligence (ASI), DePue shared.

“We’re keeping the team tight but we’re open to high-slope researchers & engineers interested in making progress in uncharted territory,” he added. As of now, OpenAI has not released additional details about this new team, and the company did not respond to a request for comment from Business Insider.

### Background on Will DePue

Born in 2003, DePue studied computer science at the University of Michigan. According to his personal website, he has “dropped out” and is currently “on leave.” DePue first joined OpenAI in July 2023, and his LinkedIn profile indicates that he has now rejoined the company. During his time at OpenAI, he contributed to training ChatGPT models and played a pivotal role in developing Sora.

Beyond his work at OpenAI, DePue has pursued several personal projects, such as a browser-based GPT implementation, a Turing-complete computer built inside Figma, and an embedding dataset project called the “Alexandria Index.”

### About Troy and Eric Luhman

While not much is publicly known about Troy and Eric Luhman, both have published research papers in machine learning. The trio—DePue, Troy, and Eric Luhman—formed the core team behind Sora, which OpenAI unveiled in February last year. Sora was publicly launched in December 2023, followed by the release of Sora 2 in September 2024.

### OpenAI’s Vision: From AGI to Superintelligence

OpenAI has long emphasized the pursuit of artificial general intelligence (AGI), the theoretical milestone when AI can reason and understand like humans. While AGI remains the company’s central mission, CEO Sam Altman signaled in a January blog post that OpenAI is already looking beyond this goal.

He described superintelligence as a future stage where AI systems would reason far beyond human capability. Altman wrote, “Superintelligent tools could massively accelerate scientific discovery and innovation well beyond what we are capable of doing on our own, and in turn massively increase abundance and prosperity.”

As DePue and his team embark on this bold new venture at OpenAI, the AI community will be watching closely for developments on the path toward superintelligence.

*Related Stories*
Business Insider continues to cover the most innovative stories shaping technology and AI advancement. Stay tuned for updates on emerging breakthroughs and the people behind them.
https://www.businessinsider.com/openai-sora-engineer-will-depue-new-team-artificial-superintelligence-2025-10

YouTube rolls out an AI feature to protect creators from deepfakes

YouTube Launches Likeness Detection Tool to Combat AI-Generated Deepfakes

As AI-generated videos, including elaborate deepfakes, become increasingly common, concerns around the misuse of personal likenesses grow. While AI has yet to solve humanity’s greatest challenges, it’s certainly raising new ones—like unauthorized use of creators’ faces in misleading content. In response, YouTube has officially launched a new tool designed to help creators protect their likenesses from AI-generated “slop.”

**What Is the YouTube Likeness Detection Tool?**

YouTube’s new AI likeness detection tool enables creators to identify AI-generated or altered videos featuring their facial likeness without permission. This feature empowers creators to manage these unauthorized uses and to request removal of the offending videos.

The goal is clear: to help creators safeguard their identities and prevent audiences from being misled by deepfake content. Currently, the tool is available exclusively to members of the YouTube Partner Program. Access will roll out gradually over the coming weeks, with all monetized creators expected to have the feature by January 2026.

**How to Use the Likeness Detection Tool**

Creators can find the new feature within the content detection tab in YouTube Studio. To start using it, creators must complete an identity verification process, which requires submitting a photo ID along with a selfie video.

Once verified, creators receive alerts whenever AI-generated videos are detected using their likeness. YouTube Studio will provide a list of these unauthorized videos, along with details such as the channel name, video title, and view count.

The tool also highlights specific segments of the videos where the creator’s likeness appears, allowing creators to review each instance before submitting a removal request.

**Early Steps in a Growing Challenge**

With apps like Sora and other deepfake technologies becoming more widespread, challenges around the misuse of identity are expected to increase. YouTube’s new tool is a positive first step toward addressing these concerns, but it’s important to note that deepfake problems aren’t limited to celebrities alone.

One ongoing concern is the data creators must submit—photo IDs and biometric information—which raises privacy questions. Many users may hesitate to provide such sensitive data, even in exchange for protection.

Looking ahead, broader measures could be beneficial, such as introducing stricter regulations on AI-generated videos or creating separate video feeds clearly marked with warnings about AI-generated content. While these won’t fully stop deepfakes, they could help reduce confusion and protect viewers from misinformation.

**Would you be willing to submit a photo ID and face scan to protect yourself from deepfakes?**

– Yes, I don’t want my face used for AI slop
– Maybe, if I can trust the company that does the detection
– No, I prefer to keep my data safe

**Stay Updated**

For the latest updates on YouTube and AI tools, subscribe to our newsletter and stay informed on how technology is shaping content creation and online safety.

**Recommended Reading**

*Iconic Phones* – Coming This Fall!

Relive the most iconic and unforgettable phones from the past 20 years with *Iconic Phones*, a beautifully illustrated coffee table book. Featuring stories of over 20 beloved devices, it takes you on a nostalgic journey through the mobile revolution that changed our world.

Sign up today to lock in your early-bird discount and don’t miss out on this stunning collection!

[Learn More and Sign Up for Early Bird Discounts Here]

*Note: Some links in this article may be affiliate links, which means we may earn a commission if you make a purchase at no extra cost to you.*

Stay connected with us for more tech news and stories on platforms like Google News.

*Recommended Stories*
– FCC Approves Cingular’s Purchase of AT&T Wireless
– Save on Verizon’s Fastest 5G Plans with Visible+ Pro: Buy 3 Months, Get 3 Free

© 2024 [Your Site Name] – All rights reserved.
https://www.phonearena.com/news/youtube-rolls-out-an-ai-feature-to-protect-creators-from-deepfakes_id175071