Who Owns Your Face in the Age of AI?

Summary

WSJ Tech News Briefing: Who Owns Your Face in the Age of AI?

Date: October 17, 2025
Host: Julie Chang (WSJ)
Guests: Jin Jo Lee (Heard on the Street columnist), Nicole Nguyen (Personal Tech columnist)

---

Episode Overview

This episode tackles two timely tech topics:

1. The speculative boom in zero-revenue energy companies, fueled by AI-driven demand for power.
2. The rising risks and ethical quandaries of AI-generated digital likeness—who owns your face as generative video technology advances?

WSJ reporters provide context, analysis, and firsthand experiences with cutting-edge tools, flagging both the investor frenzy around next-gen energy startups and the privacy implications of powerful video AI apps like OpenAI's Sora.

---

Key Discussion Points & Insights

1. The AI Gold Rush—In Energy, Not Just Tech

(00:24–03:41)

• Unprecedented Valuations for Zero-Revenue Startups

  • Oklo, a nuclear energy startup backed by Sam Altman, is valued at $25B despite no revenue, leveraging nascent technology and ambitious goals.
  • Fermi plans to build out capacity equal to all of New Mexico’s power generation, yet only has a small fraction of that secured.
  • “It's the largest market cap for a listed US Company with zero revenue.” —Jin Jo Lee (01:23)

• Investor Motivation

  • Sky-high valuations in profitable energy firms push investors toward early-stage, unproven bets with links to AI growth.
  • The “AI boom” is seen as a catalyst that could legitimize and commercialize expensive, advanced energy solutions.
  • “If there was ever a time for a new energy technology or an expensive energy technology to take off than it would be now.” —Jin Jo Lee (02:50)

• Risks of an AI Bubble Burst

  • Established, profitable companies might weather a downturn—zero-revenue newcomers risk catastrophic share price drops.
  • “Their share prices probably have the furthest to come down if the bubble bursts.” —Jin Jo Lee (03:28)

---

2. Who Owns Your Face? Dilemmas of AI-Generated Digital Avatars

(04:25–09:29)

The Rise of Sora and Other AI Video Tools

• OpenAI Sora: How It Works

  • Invite-only, iOS-available app creating rapid video avatars from short user prompts.
  • Nicole Nguyen trialed it, noting she could create a “digital avatar” after “look[ing] at a camera, speak[ing] three numbers aloud, mov[ing] my head around.” (05:32)
  • “It did look and sound a lot like me that I could put in any scenario that I could dream up.” —Nicole Nguyen (05:38)

• Believability—The Good, the Bad, and the Uncanny

  • Likeness varies: sometimes eerily accurate in detail, other times off-putting or inconsistent (e.g., odd teeth, unnatural comments).
  • “In some, my facial expressions...the texture to my skin was perfect… In others, my teeth look really off. And it said things that I would never say.” —Nicole Nguyen (06:27)

• Dangerous Flexibility and Potential for Abuse

  • Generative AI’s unpredictability—identical prompts yield different results, making control difficult.
  • The capacity to put anyone’s likeness in fabricated, contextless, or damaging scenarios raises deep concerns.

Company-Provided Privacy Protections

• Cameo Feature Limits

  • Users can see and delete any video drafts where their avatar is used—even before publishing.
  • Content featuring you cannot, in theory, be downloaded or screen-recorded.
  • “You will always have visibility into their drafts, even if they don't publish those videos. And you can delete them at any time.” —Nicole Nguyen (07:19)
  • Nevertheless, workarounds exist.

• Who Is Fair Game?—Issues of Consent and Copyright

  • Living celebrities (e.g., Taylor Swift) and major IP (like Star Wars) are blocked; historical figures such as Martin Luther King Jr. are not.
  • Nicole saw “dead celebrities” used in questionable contexts:
    • “In one of the creepiest videos that I saw, [...] [MLK Jr.] said, ‘I have a dream that Sora will change its content violation policy,’ which is something that Martin Luther King Jr. never said.” —Nicole Nguyen (07:56)

Risks for Individuals

• Loss of Control Over Likeness

  • Once images are uploaded, users often forfeit significant control, and AI tools amplify risks of manipulation and viral spread far beyond old-school Photoshop.
  • Most tools’ Terms of Service require users to have rights to any uploaded photos—but enforcement is lax and open to abuse.
  • “Once you hand your likeness over to these AI apps, you lose control of them.” —Nicole Nguyen (08:24)

• Training Data Concerns and Controls

  • Major tools let users opt out of content being used for model training, but settings require user action.
  • “OpenAI says it takes steps to protect user privacy...you can prevent the company from using your content to train its models.” —Nicole Nguyen (09:18)

---

Memorable Quotes & Notable Moments

• “If the AI bubble bursts, their share prices probably have the furthest to come down.” —Jin Jo Lee (03:28)
• “All I had to do was look at a camera, speak three numbers aloud, move my head around, and then it created this impersonation of me, this digital avatar…” —Nicole Nguyen (05:32)
• “Once you hand your likeness over to these AI apps, you lose control of them.” —Nicole Nguyen (08:24)
• “He said, ‘I have a dream that Sora will change its content violation policy,’ which is something that Martin Luther King Jr. never said, but it was definitely in his voice and in his likeness.” —Nicole Nguyen (07:59)

---

Timestamps for Important Segments

• 00:24 — Introduction to speculative energy startups and their ties to the AI boom
• 01:18–03:41 — Analysis of Oklo and Fermi valuations; risks of AI-driven investment bubbles
• 04:25 — Introduction to Sora and new AI video tools
• 05:14–07:05 — Nicole Nguyen’s firsthand review of generating digital avatars and their limitations
• 07:12–08:17 — Privacy features, restrictions, creepy AI reenactments of historical figures
• 08:21–09:29 — Broader risks: loss of control, data sharing, and model training concerns

---

Takeaways

• Generative AI is revolutionizing both energy investment and digital media—sometimes recklessly.
• Tools like Sora make it trivially easy to fabricate highly realistic fakes of anyone or anything, with safeguards still lagging behind.
• Users, lawmakers, and tech companies urgently need to clarify who controls digital likenesses and how to rein in misuse before deepfakes and unauthorized avatars become commonplace.

Transcript

[Elastic Representative]

Why do over 50% of the Fortune 500 use elastic? Because of search AI. Search AI is WatsonNext chatting one on one with your business? It's also LGCNS turning real time context into real smart cities. Find out what Elastic can do for your business at Elastic Co SearchAI welcome.

[Julie Chang]

To Tech News Briefing. It's Friday, October 17th. I'm Julie Chang for the Wall Street Journal. While some investors fret about high valuations for tech stocks, could the real bubble be in energy? We'll look into why investors are going all in on zero revenue energy companies. Plus, who owns your face in the age of AI? Our personal tech columnist joins us to discuss how our likenesses are no longer our own. But first, our Herd on the street columnist Jin Zhou Lee is following a group of energy firms, many working on small module nuclear reactors that are collectively worth more than $45 billion despite having no current revenue. Jinju joins me now to discuss. So Jinju, can you tell us a little bit about some of these energy companies?

[Jin Jo Lee]

The largest one is Oklo, which is backed by OpenAI Sam Altman and valued at about 25 billion dol billion. It's a nuclear startup that's working on a small modular reactor technology that's still pretty nascent and it's the largest market cap for a listed US Company with zero revenue. And then another large one is Fermi, also zero revenue. They have these very ambitious plans to build out 11 gigawatts worth of power. That's about the amount of capacity that the state of New Mexico has in Texas to power data center customers. They have a few natural gas turbines lined up, but it's only enough to meet like 5% of their total goal. And neither company has any binding customer agreements.

[Julie Chang]

So why are people investing in these more speculative energy companies?

[Jin Jo Lee]

Well, there's generally a lot of interest in any company that has potential to profit from the AI boom. And some of the already profitable energy companies are already very pricey. Some of their stocks are trading at like 100 times forward earnings. And I guess there is a case to be made that if there was ever a time for a new energy technology or an expensive energy technology to take off than it would be now when there's so much sudden need for power from some very well capitalized tech companies.

[Julie Chang]

So then on the flip side, what happens if these energy companies don't take off because say the AI bubble bursts?

[Jin Jo Lee]

Yeah, I mean if you are a tech company that is already profitable and the AI bubble bursts, then you're still probably going to be fine, right? Maybe your revenue projections out in the future are not as big as you had hoped, but you'll still survive. The issue with some of these energy companies is that they're not backed by any revenue or are loss making, so their share prices probably have the furthest to come down if the bubble bursts.

[Julie Chang]

That was WSJ Heard on the street columnist Jin Jo Lee. Coming up, your digital doppelganger is mere clicks away thanks to AI. But what if you don't want it to be? That's after the break.

[Elastic Representative]

Why do over 50% of the Fortune 500 use elastic? Because of search AI. Search AI is Watsonnex chatting one on one with your business? It's also LGCNS turning real time context into real smart cities. Find out what Elastic can do for your business at Elastic co Searchai.

[Julie Chang]

AI Videos are taking over social media. You may have already spotted the Sora logo watermarked on videos as you scroll through Instagram or TikTok made by OpenAI. Sora can take a genuine video or photo of someone and spin up a virtual lookalike in mere minutes. And it's not alone, but Google has its Nano Banana Image generator and Apple's Got Image Playground. WSJ personal tech columnist Nicole Nguyen is here to discuss these tools, the potential risks they pose, and the safeguards that are in place. And a quick note. News Corp, the owner of the Wall Street Journal, has a content licensing partnership with OpenAI. So Nicole, the big video generating app right now is OpenAI's Sora. Tell us about Sora. Where can I find it? How does it work?

[Nicole Nguyen]

So Sora is currently invite only. It's available to anyone to Download via the iOS app store. But to be able to use Sora, you need an invite code. It is built on OpenAI's latest video generation model and the reason why it's so buzzed about is because you can create your own digital likeness within minutes. All I had to do was look at a camera, speak three numbers aloud, move my head around, and then it created this impersonation of me, this digital avatar that did look and sound a lot like me that I could put in any scenario that I could dream up and I could also allow other people who are on the platform to create videos with my digital avatar appearing in them and also open it to everyone, which I definitely did not because that would be crazy.

[Julie Chang]

Okay, so you tried out that digital doppelganger or cameo feature. What did the results look like? Were they pretty believable?

[Nicole Nguyen]

So I would say mixed. I Mean, at first, it really depended on the scenario of the video. And the weirdest thing about Generative AI is the same prompt will produce a different response every time. So the likeness does not look exactly the same in every single video. In some, my facial expressions, the flyaways in my hair, the texture to my skin was perfect on point. And then in others, my teeth look really off. And it said things that I would never say. And it didn't look like me. But I was definitely creeped out at its capability in, like, the best circumstance, with the most finely tuned prompt that it could create something that looked and sounded just like me and make it do something that I have never done or did say in real life.

[Julie Chang]

And are there any restrictions on using Sora at the moment? Did the companies say what kinds of privacy protections there are for users?

[Nicole Nguyen]

There are elements of this cameo feature that are really elegantly designed. So, for example, if you let other people use your cameo in their videos, you will always have visibility into their drafts, even if they don't publish those videos. And you can delete them at any time. And those videos that are created by them and feature, you cannot be downloaded or screen recorded. And I'm sure that there are ways around that, but the baseline protection is pretty good. I did, however, see a lot of dead celebrities in the app. There are restrictions around living celebrities, like, anything that mentions Taylor Swift or intellectual property like Star wars will be immediately blocked. But Martin Luther King Jr. For example, was fair game. And in one of the creepiest videos that I saw, it looked like this grainy footage, you know, historical footage that looks genuine. And he said, I have a dream that Sora will change its content violation policy, which is something that Martin Luther King Jr. Never said, but it was definitely in his voice and in his likeness.

[Julie Chang]

So what are some of the risks involved with using this tech?

[Nicole Nguyen]

The main risk is that once you hand your likeness over to these AI apps, you lose control of them. We've known since the dawn of the Internet, once you post a photo, it can be Photoshopped endlessly. It can be shared and distributed in many ways without your consent. But these AI tools really supercharge both the the manipulation and distribution of these images. And a lot of platforms. ChatGPT, Google's Gemini, say you need to make sure you have the rights and permission to upload a photo that you're going to manipulate. But just because they say that doesn't mean that people do that. And so anyone could really upload an image featuring you to these tools and create something that looks like you, and now those companies have access to your face as a part of their training data. OpenAI says it takes steps to protect user privacy. And in the app settings, for example, you can prevent the company from using your content to train its models. And that's true of other AI apps, too.

[Julie Chang]

That was WSJ personal tech columnist Nicole Nguyen. And that's it for Tech News Briefing. Today's show was produced by me, Julie Chang. Jessica Fenton and Michael Lavalle wrote our theme music. Our development producer is Aisha Al Muslim, Chris Zinsley is the deputy editor and Falana Patterson is the Wall Street Journal's head of news audio. We'll be back later this morning with TNB Tech Minute. Thanks for listening.

[Elastic Representative]

Why do over 50% of the Fortune 500 use elastic? Because of search AI. Search AI is what's next, chatting one on one with your business. It's also LGCNS turning real time context into real smart cities. Find out what Elastic can do for your business at Elastic Co searchai.