Michael Schumacher's family take legal action over AI-generated interview | Hacker News Reader

Michael Schumacher's family take legal action over AI-generated interview

skysports.com

76 pointsignoramous33 comments

I don't think the AI part really matters here. If the magazine had instead just written his answers themselves, would the legal questions be any different?

As I understand it, what they did was go to an online chatbot (probably character.ai), conduct the fake interview, publish it with a headline and cover that suggested it was real, and then admit only at the end of the article that it was from an AI website.

Presumably, the AI in question was just a generic LLM prompted to act like Schumacher - I would be surprised if it were specially trained on his words or anything. It feels like the only legal question is whether it was fraudulent to portray the interview as real, or whether the admission at the end is sufficient cover. I don't think this case would give us any real insight into the legal issues surrounding AI in particular.

It probably wasn't AI at all. That magazine has just about 25,000 subscribers, so you can guess the quality of their journalism. Out of their authors most likelike just made it up and put the "AI" label on it to make it "more exciting".

The BBC News report (https://www.bbc.co.uk/sport/formula1/65333115) says that it was character.ai. They don't say how they know that though (presumably it's disclosed in the magazine itself?)

> presumably it's disclosed in the magazine itself?

Yes, indirectly, they have screenshots of the answers with a blue "c.ai" box in them.

Wouldn't an AI generated interview be cheaper than paying an author to make one up?

Good argument, didn't think of that. And there may not even be a journalist working anymore at that magazine.

> That magazine has just about 25,000 subscribers, so you can guess the quality of their journalism.

Not saying anything about this particular magazine, but in general, I don't think that you can judge the quality of a publication's journalism by their subscriber count.

> publish it with a headline and cover that suggested it was real

The subheading is "It sounds almost real! +++ What's behind this" ("Es klingt täuschend echt! +++ Was dahinter steckt"), allowing them to weasel themselves out of just flat out lying. That's the schtick of "Die Aktuelle".

On the actual interview page, they explicitly disclose that it's not a real interview in their, eh, unique way of writing:

Talk to him just once. Ask him how he is really doing. And almost ten years after his tragic finally get some answers! Not meager, nebulous half-sentences from friends. But answers from him! From Michael Schumacher, 54! Here it is - the incredible interview!

With redemptive answers to the most burning questions that the whole world has been asking for so long. How is he doing since the tragic accident in 2013? The answer: "My life changed completely after that. It was a terrible time for my wife, my children ... I was so injured so badly that I was in a kind of in a kind of artificial coma for months because my body wouldn't have been able to take it all."

But the most important question: how does he feel today? The answer gives hope: "Much much better than years ago. With the help of my team, I can I can even stand on my own again and even slowly take a few steps and walk ... My wife and my children have been a blessing to me and without them I would not have made it. Of course, they are also very sad about how everything has gone ... They support me and stand firmly by my side."

When asked about whether he follows the careers of Gina, 26, and Mick, 24, the answer is: "Of course! I follow both of them with great pleasure ... They really make me and the whole family very proud."

And on a possible possible appearance in the future: "I have to be very cautious about it, because I don't want to let I don't want people to think too much wrong about me."

Nothing wrong? Did Michael Schumacher really say everything like that himself?

The Interview was on the Internet. On a site that has something to do with artificial intelligence, or AI for short. What is behind all this? DIE AKTUELLE went in search of clues. There are actually Internet sites where you can talk to celebrities. But the answers are provided by artificial intelligence.

But how does this AI know the personal backgrounds? About marriage, children and illnesses? Someone must have entered the information - like in Wikipedia - on the Internet! Was it really Schumi himself who the information from his sickbed? From his sickbed? Or someone from the family, nurses or employees?

The answers sound deceptively real! Too good to be true? If Michael Schumacher would speak to us - it would be a world sensation. And a miracle, also through our prayers. A real miracle.

(Yes, that's the entire interview on 2 A4 pages. That's it. They asked 4 questions to a character.ai bot. The magazine has many pictures, little text.)

>fraudulent to portray the interview as real

how is that different to the reporting on the Oasis lost tapes made by AI? Where they first exclaim "Oasis is back! Except, not really." [1][2]

Seems it depends how far you read into any story and how deep they bury the lede. Of course the wrag magazine was more egregious in their presentation elsewhere but other than the tastelessness of it all I don't see much different than printing "what would jesus do about your horoscope? NO REALLY! we asked him"

whereas with Schumacher they are tongue-in-cheek conflating the idea that AI chatbots are uploaded consciousnesses and so desirable quote factories for their dwindling empire.

[1] https://decrypt.co/137160/ai-oasis-aisis-breezer-liam-gallag...

[2] https://www.stereogum.com/2220920/oasis-ai-album-liam-gallag...

There's nothing "AI specific" in this, it's just shady journalists making up stuff to lure readers.

I agree that it's mostly the same problem...presenting something that wasn't an interview with him as being an interview with him, regardless of how it was done. But other sources do mention they used Character.ai to generate the interview answers.

Not the technology itself maybe, but it does seems relevant to state of the business environment to point out this fraud was performed largely by using a service created for practically this use case. Sure, they could have made up the interview from scratch, but this company obviously provided at least a valuable facilitation, and arguably the AI hype is the reason anyone bothered with executing the idea in the first place.

Character.ai claiming they didn't want this sort of stuff is like early youtube insisting they didn't want people to upload clips from the Simpsons.

How much money one has to spent, how large must be entourage of lawyers and lobbyists to achieve this level of aggressive privacy? The other example is Schwartz family of Lidl and Kaufland, and probably countless industrialists and aristocrats living in DACH countries. You mention them and wait for cease and desist letter in your mailbox, that's how they maintain their crystal clear image or no image at all.

Sounds like a much better strategy than annoying us all with media adulation. Maybe there are some positives to going that length for the sake of privacy.

If he was healthy don't worry, we would see his face on every billboard, newspaper, TV ad, and other formats advertising caffeine drinks, car brands, electronics, watches or whatnot. We'd all forcefully learn about his new lover, yacht, or car. But well things went south... "now I'm not a celebrity!".

Not that much actually and it does not cost that much especially when there really has been a misuse of likeness. The same crawlers that make search work make it easier to uncover potential likeness and trademark misuse.

In my opinion with the recent Ice Spice song and AI generated Drake, I feel like an artists voice and likeness should be covered by copyright laws. It’s kind of absurd that someone could make a song with someone else’s voice, and not pay royalties of any sort.

David Guetta released a song a few months ago with an AI sample, so he may be the pioneer of this new era. He said he used an online text generator to create an Eminem rap verse. He then copied the text and pasted into another website to generate Eminem’s voice. He said he did it to just mess around with AI and being creative, but he didn’t agree that it should be legal for him to do that.

https://youtu.be/lm-o6joqChc

>In my opinion with the recent Ice Spice song and AI generated Drake, I feel like an artists voice and likeness should be covered by copyright laws. It’s kind of absurd that someone could make a song with someone else’s voice, and not pay royalties of any sort.

It's absurd that someone could think it's fair to send men with guns to make someone pay them just for copying how they speak/sing. And unprecedented; imagine if Elvis' estate was allowed to go after every Elvis impersonator who sung non-Elvis songs in Elvis' accent!

Impersonation is a bit different.

I wasn’t thinking about men with guns. I was referring to the current music copyright and royalty laws that are upheld by groups like RIAA.

You can legally sample another artists music, even without their permission or knowing. However, they have to be given credit, and everyone inline that was involved with the original song composition get a royalty.

An example of this was Jay-z’s 99 problems. Somewhere in the 90’s, Too Short mentioned to Ice T that “he had 99 problems, but a B wasn’t one”.. Ice T thought that was slick and wrote a song.

A couple years later, Rick Rubin and Jay-z borrowed that hook and put out a hit. Ice T said that he didn’t know they had recorded the song, and he only found out when he heard it on the radio. He said it was all cool though because he received writing credits, and royalty payments as a result.

What about every impersonator? What if you just happen to sound like Drake? Who owns that voice?

Cover bands and parodies are already covered by current copyright laws.

This is a bit different, and is a bit uncharted.

People magazine with their celeb news - is it any different ???.

The Schumacher family esp his wife is very cagey of the true status of Michael today.

They could have handled it better IMHO - he was a very public figure and people are naturally curious - they simply could have ended all the speculation by stating that he is no longer the person we knew as he suffers from acute brain trauma and is no longer in a condition to engage with the public.

It seems like someone was just bored waiting for the Baku GP

There's going to be a lot of these frivolous lawsuits as lawyers try to extract cash from the ignorant and gullible before people understand what deep fakes are. I only hope that the judges in these cases are technically informed enough to not set stupid case precedent.

In which world should fabricating an interview with someone be anything other than fraud and potentially libel?

The stupid precedent is "you can say that anyone said anything (even damaging things to them) without facing any consequences".

It would only be fraud if they claim it was real. But those magazines are masters in deceiving their lazy readers. They likely mentioned somewhere that it's not real, to avoid the law.

> In which world should fabricating an interview with someone be anything other than fraud and potentially libel?

https://en.wikipedia.org/wiki/Hustler_Magazine_v._Falwell

This doesn't have anything to do with deep fakes.

My understanding is the article says it’s an interview with him, but it’s not an interview with him…

If you’re saying that AI can represent real humans, then than is a whole can of worms that society can’t support.

Sounds similar to crypto-talk: "but you don't understand, it's based on blockchain". It doesn't matter what technology is used, but what action is done.

This has got nothing to do with the technology. If you sell an "interview with Michael Schumacher" but it's actually your buddy in a boiler suit instead, you have probably committed a fraud. The tech here is just enabling the fraud to be more convincing.

> The tech here is just enabling the fraud to be more convincing.

So it does have to do with the technology?

I'm saying the tech is not germane to the question of whether or not this was deceptive or illegal, the tech amplifies the effect.

Reply on news.ycombinator.com