Posted on 07/28/2025 5:49:30 AM PDT by Openurmind
OpenAI CEO Sam Altman is warning of an "impending fraud crisis" in which practically anybody will be able to imitate not only the voice but even the likeness of others, using the type of AI that his company is helping pioneer.
During a chat with Federal Reserve vice chair Michelle Bowman in Washington, DC, Altman argued that the banking industry must modernize to avoid having its customers fall victim to widespread fraud.
"I am very nervous about this," he told Bowman. "A thing that terrifies me is apparently there are still some financial institutions that will accept a voice print as authentication for you to move a lot of money or do something else. You say a challenge phrase and they just do it."
"AI has fully defeated most of the ways that people authenticate currently, other than passwords," he went on. "Right now, it’s a voice call; soon it’s going be a video or FaceTime that’s indistinguishable from reality."
Excellent ideas, both of you.
Carry_Okie suggests building security (basically SSH certificates) into all future cameras / microphones / etc. While a good idea, I suspect this wouldn’t be practical and wouldn’t solve the problem of existing hardware.
Bob434 postulates a passphrase. This could work well.
I already offered 2-factor, either phone or authenticator app (as goodnesswins suggests).
I think it’s Trouble - spelled with a capital “T” that rhymes with ????? In any case, it’s TROUBLE!
Wouldn't you miss that 30 minute hold on customer service line (which repeats that "Your call is important to us")? And you have to press 1 for English only to get someone who barely speaks it with a heavy accent. You'd miss that too,wouldn't you?
The real danger is with political stuff close to an election, where fake, AI-generated "video" of a candidate saying something politically damaging starts making the rounds. Or other fake-AI people saying they saw/heard the candidate doing something politically damaging.
After the fact lawsuits for defamation or whatever that wouldn't be resolved before the election to provide clarity don't address that problem. Just a really dangerous phenomenon in a republic.
They are already using facial recognition, fingerprints, and voice recognition as methods to unlock smart phones. It would not be hard to use this tech to created a unique one off hashed encrypted token that could be shared only with the important sites like banking and medical portals. But I would not want it connected to my real person for any other accounts anywhere else. So I would have to have two, one used as an anonymous token. But if someone stole your phone...
I didn't know what that meant so I asked my wife if it was something to worry about. She said no, if the on/off switch don't work just unplug it.
“Frank Sinatra doing Led Zeppelin would be awesome!”
Not the Chairman of the Board, but here’s an up-and-coming country group doing Led Zeppelin:
https://youtu.be/l5_9fYU8jd4?feature=shared
I can’t wait to see them in concert! /s
” The real danger is with political stuff close to an election, where fake, AI-generated “video” of a candidate saying something politically damaging starts making the rounds. Or other fake-AI people saying they saw/heard the candidate doing something politically damaging.
After the fact lawsuits for defamation or whatever that wouldn’t be resolved before the election to provide clarity don’t address that problem. Just a really dangerous phenomenon in a republic.”
Even false evidence and crimes could be created against you... “Security Cameras recorded the crime and identified the suspect as it happened”. They could create all kinds of serious fake crimes and evidence with your face on it.
This AI is a huge problem that folks just do not realize it yet.
I had a talk with a friend of mine, who is working on a business venture. He said a new exchange method is in the works, uses a blockchain like technique to verify the money is real, and makes transaction costs less than a penny.
It could help with some of this.
It could make debit cards obsolete. All digital, of course.
It is said to be much, much cheaper than the current banking method of moving money.
“Playing with fire” might be another way of putting it.
I avoid AI like I avoid Facebook.
“I agree. A “Pandora’s box” has been opened and AI has jumped out. Maybe that’s a little TOO alarmist.”
Absolutely not, you are right on the money. It is an extremely serious problem... The most dangerous thing humanity has ever faced. They are going to enslave us with this technology...
The present standard anti-fraud protection is to send a code by cell phone. It works, unless someone steals the cell phone.
You can avoid AI, but AI won’t avoid you.
He just figured this out? He’s on the ball.
“The internet is going to be full of even more garbage than usual”
What is sad is the number of gullible people, (I include myself with that label at times), who will fall for manipulated videos and pictures that are CLEARLY photoshopped or altered in some way.
What world will we be in when we cannot discern real videos and pictured with those generated by AI?
Pick your favorite, or least favorite, politician in a video molesting a child or in the act of a violent rape going viral around the world, before it is disproven.
Now picture the release of something like this 5 days before an election. It’s coming!
“ Wouldn’t it be great if society went back to face-to-face business deals where actually knowing someone as a trustworthy person mattered? ”
That would be great, but most people dont live like that anymore
This is becoming very obvious.
A few weeks ago I asked ChatGPT “how many seconds of voice sample” is needed to construct a good enough model to fool most Voice ID schemes? The answer that came back was “3 to 5 seconds,” although there are countermeasures that can be taken to make voice fakingmore difficult. So, the obvious conclusion is that you can’t talk to anyone, and you’ve got to use a synthesized voice answering message as this unfolds. Your very simple “Not here, leave a message at the beep” could become a part of identity theft.
So, yesterday I had another idea. Someone will start licensing the vocal tract models of popular actors and singers. Want ChatGPT (or the others) to talk to you like your favorite hot actor/actress? There will be voice model available for that. Actors and singers will have their vocal tract parameters copyrighted (why give it away for free?).
There are already fake AI bands, Rick Beato talked about this.
https://www.youtube.com/watch?v=3Nlb-m_vKYM
To me, the danger with political stuff is that the window to discover and correct it is so small. Also, many voters may only see the initial fake video, and not be aware of the later proof that it was fake. Or they may not believe that it was fakes, especially if it is something negative about the opposing candidate.
I believe Texas is passing a law to criminalize AI-deep fakes that portray a candidate saying or doing something they did not say or do.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.