Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

They stole my voice with AI

jeffgeerling.com

Post reply on HN
Print view View original

391–400 of 461 posts

Re: They stole my voice with AI

#391

Post by disqard »

marginalia_nu wrote: post #182
sureglymop wrote: post #171

Earlier quoted context omitted.

My specific fear is that if a picture of you next to your name is available online, that becomes part of the training set of a future model. Paranoically, I do not have any picture of myself available online. I could then trivially generate pictures or even videos of you e.g. by knowing your name. Of course that's just an example but I do think that's where we are headed and so the concept of "trust" will change a lo…

Seems like the end game for that technological development is kind of self-defeating. Once it's 2 clicks away to generate a believable video of someone at the kkk kitten barbecue getting along with ted bundy and jeff epstein, surely the evidence value of that would dwindle, and the brief period in history when video evidence was both accessible and somewhat believable will come to an end.

You're right -- there's nothing axiomatic about the trustworthiness of pixels, and there was a brief window of Time in human history when photos/videos had the qualities you mentioned, and (sadly) that era will end soon.

Being savvy means not treating photos/videos as "untrustworthy", which (ironically) is one of the things the Western world made fun of the Taliban for -- their fundamentalist view called it "haram":

https://www.deseret.com/1997/10/6/19338048/taliban-army-bans...

I worry about written matter too, and here as well, vast masses of LLM-spewed text will surely pollute the landscape irreversibly.

Re: They stole my voice with AI

#392

Post by disqard »

fennecbutt wrote: post #301
ryzvonusef wrote: post #135

Everyone has their own fears about AI, but my fears are especially chilling; what if AI was used to imitate a person saying something blasphemeous? My country is already has blasphemy lynching mobs based on the slightest perceived insult, real or imagined. They will mob you, lynch you, burn your corpse, then distribute sweets while you family hide and issue video messages denouncing you and forgiving the mob. And thi…

The people are the problem not the tool.

I'm reminded of Chris Rock's "Guns, don't kill people, bullets do!"

In a more serious vein, this is definitely about unleashing an extremely powerful technology, at scale, for profit, and with insufficient safeguards (imagine if you could homebrew nuclear weapons -- that's inconceivable!)

There will be collateral damage. How much, and at what point will it trigger some legislation? Time will tell.

Re: They stole my voice with AI

#393

Post by Corrado »

So, do we "solve" this by impersonating a voice from a very litigious company? Let's say we clone Micky Mouse's voice to say some blasphemous things. Does that bring the full weight of Disney down on this? Will that force a change? Would it be a good change?

Re: They stole my voice with AI

#394

Post by rldjbpin »

from the discourse here, the main pain point really is the accessibility to do something like this thanks to the new models.

IANAL and not sure about regional precedence on these topics, but there are plenty of ads where lookalikes or voice actors are used to use someone's likeness. they are mostly in satire, but there is yet to be a case where there was a litigation over this or prior approval needed.

we have ai-based voice abuse in the political sphere, and where there was only one legislation for banning the use in voice calls for one country (https://news.ycombinator.com/item?id=39304736), another country actively used the same underlying tech to aid their own rallies (https://news.ycombinator.com/item?id=40532157).

the tools are here to stay, but what is fair use needs to be defined more than ever.

Re: They stole my voice with AI

#395

Post by pjc50 »

rustcleaner wrote: post #329
pjc50 wrote: post #193

Earlier quoted context omitted.

The US equivalent is much less labour intensive than a lynch mob: it's mass shooters radicalized by things they've read on the internet. Or https://www.npr.org/2024/09/19/nx-s1-5114047/springfield-ohi... , where repeating racial libel causes a public safety problem. While this kind of incitement in no way requires AI, it's certainly something that's easier to do when you can fake evidence. See also https://www.bbc.co…

>radicalized by things they've read on the internet That take pretty much throws the shooters' preceding years of anguish under the bus to make a political score. "Ban the wrongspeech!" is a seductively easier proposition than ending youth bullying and ostracism.

Quite a few (not all, but the more famous ones) write manifestoes in which they describe how they were radicalized, cite other manifestoes and mainstream racist commentators who inspired them, along with Adolf Hitler, and make it very clear what their motives are.

> youth bullying and ostracism

... which itself counts as free speech?

A lot of people get bullied in a lot of countries. Few of them turn to mass murder. The mass shooter phenomenon is mostly America, plus a few people who've been radicalized by the same racist discourse on the same message boards (Anders Brevik, NZ mosque shooter).

In quite a lot of cases closer investigation suggests that they were ostracized precisely because they'd already chosen hate.

Re: They stole my voice with AI

#396

Post by pjc50 »

datavirtue wrote: post #297
pjc50 wrote: post #193

Earlier quoted context omitted.

The US equivalent is much less labour intensive than a lynch mob: it's mass shooters radicalized by things they've read on the internet. Or https://www.npr.org/2024/09/19/nx-s1-5114047/springfield-ohi... , where repeating racial libel causes a public safety problem. While this kind of incitement in no way requires AI, it's certainly something that's easier to do when you can fake evidence. See also https://www.bbc.co…

People don't shoot up crowds because of something they read on the internet. They do it because they are done with life. Everything after that is retrospective reasoning.

Some of them (e.g. Breivik) write manifestoes in which they list all the stuff they read on the internet that motivated them.

Re: They stole my voice with AI

#397

Post by adityaathalye »

chefandy wrote: post #127
scotty79 wrote: post #112

Earlier quoted context omitted.

> feels like the wild wild west of intellectual property and copyright law Copyright seems to always have one or another wild wild west going on. Maybe you are in the wrong place if the world constantly jumps and kicks from under you trying to throw you off?

Anyone that thinks this is completely untrodden ground for copyright should ask an expert to definitely determine if someone's use of something is covered under fair use if it doesn't exactly and clearly satisfy all of the test prongs.

GP here. I agree. By "lack of precedent" I meant lack of precedent specific to the implications of gen AI's use of data scraped indiscriminately, without attribution or consideration. As far as I can tell, the various cases in progress will (eventually) set some precedent about fair use in this specific context. LLM-owning companies are already hedging future liability by paying for content and/or doing the usual terms of use bait-and-switch (oh yeah, and when you click "okay", you agree to let us use all your data to train our AI models).

Re: They stole my voice with AI

#398

Post by lesostep »

oehpr wrote: post #370

I want to ask and answer two of my own questions here: 1. Why clone Jeff's voice? When I was messing with stable diffusion using Automatic1111's interface, I noticed it came with a big list of artists to add to the prompt to stylize the image in some way. There was a big row in the media about ai art reproducing artists work and many artists came forward feeling it was a personal attack. But... I mean the truth is mo…

>> But... I mean the truth is more general than that. When I pressed a button to insert a random name into a prompt, my goal was not "yes give me this person's art for free", it was "style this somehow".

yeah and that's the problem. The style of an artist is a developed thing. To think that one could borrow your style not through learning and caring, but through mathematically analyzing the width, and colors, and patterns and applying it to a random noise — that's kinda insulting. If nobody cares about my real work, why do they care about using my style, then? Develop your own an teach your AI on that, if there really isn't any difference.

People say that AI learns how a human would. But a human wouldn't (couldn't!) learn like an AI can. He can't look at the pixels, can't mechanically churn through patterns. If someone can learn from art like AI learns art, I would also be opposed to them learning anything from me :D

Re: They stole my voice with AI

#399

Post by nh2 »

eth0up wrote: post #271
nh2 wrote: post #263

Earlier quoted context omitted.

Ask your cousin to visit her and video call you together? Or go there for a weekend and check?

If my cousin (in US too) was equally skeptical, I'd contact either the US embassy or Ecuador embassy and arrange a wellness check with local LE. But while that offer and others stand, that's not my jurisdiction presently. Regarding travel resources, I live well enough but have to look steeply upward to see the poverty line, negating such options. For now, I'm satisfied being a fool.

Ah, I misunderstood, thinking that your cousin lives close to your aunt. I was suggesting that somebody physically close could do the video call.
Post reply on HN