0:00

Hey there cyber listeners Matthew here this episode

0:02

was brought to you by delete me a

0:05

few years ago. I did some reporting that uh Made

0:08

quite a few people pretty angry I got a

0:10

few death threats and I started to worry about

0:12

how much my private information was available in public

0:14

so we've got a safety team here at vice

0:16

and that helps journalists navigate these kinds of situations

0:18

and The single best thing

0:21

they did was sign me up for delete me

0:24

So, you know, how would you like to wake up one

0:26

day and discover your bank account has been emptied or

0:28

get an overdue notices? For credit cards you never applied

0:31

for or worse You don't

0:33

realize how much of your information is available to

0:35

scammers on the internet and how susceptible you are

0:37

and your family is to

0:40

identity theft and fraud Now

0:42

you can get your data removed with

0:44

delete me These are just two of

0:46

the many reasons that I personally recommend

0:48

delete me delete me is a subscription

0:50

service that removes your personal Info from

0:52

the largest people search databases

0:54

on the web and in

0:56

the process helps prevent potential ID theft Doxing

0:58

and phishing scams and this is not just

1:01

a one-time thing either. It is always working

1:03

for you I've been using

1:05

it for a few months now and

1:07

I'm constantly getting updates about

1:10

where my information is popping up

1:12

how to remove it and One

1:15

of the best parts a lot of the stuff just

1:17

gets automatically removed for me to put

1:19

it simply delete me does all the hard Work of

1:21

wiping you and your family's personal info off

1:23

of the web data brokers hate delete me

1:26

when you sign up for delete me It

1:28

immediately goes to work scrubbing your all your

1:30

personal information from data broker platforms When

1:33

you sign up delete me immediately goes

1:35

to work scrubbing all your personal information

1:37

from data broker platforms Your

1:40

personal profile is no longer theirs to

1:42

sell So take control of

1:44

your data and keep your private life private

1:46

by signing up for delete me now

1:49

It is special discount for our

1:51

listeners today get 20% off

1:53

your delete me plan when you go to join delete

1:56

me Dot-com slash cyber and use

1:58

promo code cyber at check checkout.

2:00

The only way to

2:03

get 20% off is

2:05

to go to joindeleteme.com/cyber,

2:07

enter code cyber at

2:10

checkout. That's joindeleteme.com/cyber code

2:12

cyber. Hey there cyber listeners,

2:14

Matthew here. Just a quick show note

2:17

at the beginning of the episode, we

2:19

get into legislation around artificial intelligence in

2:22

this episode. It's a really great

2:24

conversation. I want to note that

2:26

hours after we got done recording

2:28

this call, the Federal Communications Commission

2:30

declared that the use of AI

2:32

generated voices in robocalls is a

2:35

scam and is illegal. The

2:37

way they did this was pretty interesting because

2:39

it was not new laws.

2:41

It was using old

2:43

laws to basically confirm that this

2:45

new technology falls under the auspices

2:48

of a scam. It

2:50

was essentially a ruling that used

2:52

the Telephone Consumer Protection Act, which is from 1991

2:55

to say that

2:59

generating a fake voice

3:02

and calling someone and trying to get them to do

3:04

something is a scam. As

3:06

we talk about in this episode, this happened

3:08

because before the New Hampshire primary voters were

3:11

getting calls from an AI generated Biden who

3:13

was telling them not to vote. Again,

3:16

this happened literally hours after

3:18

we got done having this conversation and

3:21

that is why it is not mentioned in this

3:23

episode. So

3:54

we are with Leah and Janice.

3:56

Can you all introduce yourselves? Janice,

3:58

will you start? Hi,

4:01

I'm Machana Thros. I am a

4:03

senior editor at Motherboard where I cover AI

4:05

and other topics. I'm

4:09

Leah Holland. I use their sheep pronouns and I'm

4:11

the Campaigns and Cons Director at Digital Rights

4:13

Organization, FACT for the Future. And

4:16

can you tell me a little bit about what FACT

4:18

for the Future does and what you all are working

4:20

on? Yeah, so

4:22

FACT for the Future is a

4:25

queer women-led national digital rights

4:27

nonprofit that really focuses on

4:29

issues of surveillance and privacy

4:31

and censorship. We like to

4:33

be the grassroots voice of

4:36

the users of technology and

4:38

speak up for their

4:40

interests when corporations or legislators

4:42

are making bad decisions on

4:44

behalf of the masses online.

4:47

So Janice, you've been talking about doing

4:50

kind of an AI episode

4:52

with this specific event for

4:54

a while. And then we had a

4:56

news story. There's a pretty great news peg for it. I

4:58

mean, it's a horrible story, but it's a good news peg

5:00

for what we're going to be talking about. I wanted you

5:03

to kind of run us through that first. It's kind of

5:05

like table setting. Yeah, so I

5:07

think it was last week. There was

5:09

this big story about Taylor Swift, as

5:11

a lot of big stories tend to

5:13

be these days. And

5:15

essentially, there was

5:17

this website that started producing

5:20

deep fake AI nudes of

5:22

Taylor Swift. And they

5:24

started, of course, spreading on social

5:27

media, including on Twitter, which is

5:29

now called X because Elon Musk

5:31

is an edgelord and

5:34

Facebook and Instagram and other

5:36

places and Reddit. And

5:38

basically, this was entirely

5:40

a thing that was entirely predictable to pretty

5:42

much anyone who has been paying attention to

5:45

the space. These are

5:47

AI models that are able to

5:49

generate images and they've been getting

5:51

increasingly sophisticated, increasingly powerful. We've

5:54

been covering them for quite a while at motherboard

5:56

since the days when they

5:58

released Dali, which is opening eyes

6:00

to the first foray

6:03

into these kinds of like

6:05

generative AI things. Yeah,

6:07

it kind of ended up

6:09

in this kind of big headline

6:11

way where, you know, here's a

6:13

famous woman who is essentially being

6:16

humiliated on a massive, you

6:19

know, international scale by this

6:21

technology that is created by

6:24

large tech corporations that are

6:26

for the most part completely

6:28

unregulated from producing these tools.

6:31

And this is a thing that usually, as you

6:34

said, we've been covering this for a long time, this has been

6:36

going on for a while, years. And it seems to

6:38

me that a big part of the

6:40

news story this time was that it kind

6:43

of breached containment, right? It

6:45

went from discord groups where

6:47

it's being shared privately, which doesn't make it

6:49

right. I'm just saying that that's kind of

6:51

where these things kind of percolate in an

6:54

unfortunate way and then made it into more

6:56

mainstream sources. Do we have any

6:58

idea like how or why that happened? Like

7:00

why did this start getting circulated on Twitter

7:02

all of a sudden? Yeah,

7:05

well, I mean, there's been a couple of events leading

7:07

up to this that have been kind of testing

7:09

the waters, so to speak. And

7:11

like there were a couple of incidents

7:14

that were, you know, cited in some

7:16

of this legislation that we're going to

7:18

talk about that involved, for example, people

7:20

making fake songs

7:23

that were essentially used

7:25

like a large cache

7:27

of audio of the singing voices of

7:29

people like Drake and The Weeknd in

7:32

order to produce new songs that

7:34

sounded like they were being sung by

7:36

those artists. And

7:38

that was like one thing that happened. That was, I

7:40

think, a couple of months ago. And

7:43

then, you know, there was, of course,

7:45

that viral George Carlin

7:47

standup routine that was awful, which

7:49

we wrote about where the

7:51

George Carlin family is actually suing

7:53

the tech

7:56

podcast, which is like sponsored

7:58

by an A.I. company. that

8:01

is completely anonymous, which is really weird

8:03

and shady. And they won't, the

8:05

comedians have signed NDAs and won't disclose the

8:07

exact nature of their relationship or who's behind

8:09

the AI. Also, fun

8:11

follow up, after the lawsuit dropped, one of

8:13

the guys copped to writing the whole thing.

8:16

Yeah, I saw that.

8:18

Which is really fun, which is like another, like put a

8:21

pin in that. We'll

8:23

come back to that. I don't know if that's gonna

8:25

really save you, like... No, it's not. I

8:27

think, by saying, oh yeah, by the way,

8:29

I wrote it. Ha ha, just kidding. It wasn't

8:32

an AI. Yeah, I don't know if that's gonna

8:34

really help your gaze. No,

8:36

absolutely not. So, Leah,

8:38

what did you, I assume you were following

8:40

the story, what did you make of it? And

8:43

also, what did you make of the calls

8:45

for legislation afterward? Because there

8:47

was quite a bit of them. Yeah,

8:49

I made of it

8:51

that something like this was more

8:54

or less inevitable. And the impacts

8:57

of AI impersonation or

8:59

deep fakes or humiliation

9:02

of women

9:04

and people from marginalized communities in particular is

9:06

only gonna get worse

9:10

as we adjust the new reality of these

9:12

tools being so broadly available

9:15

and don't really have... If we even have

9:17

on the internet at all, maybe cultural worries

9:19

about what we're supposed to do or aren't

9:21

supposed to be doing with images and what

9:24

have you, I think that this is

9:26

something that we are right now

9:29

in the start of adjusting to

9:31

both when it comes to three

9:35

different categories of people that I think

9:37

are important to think about. The first

9:39

of which is your Taylor Swift, the

9:41

second of which is your abuse survivors,

9:44

your everyday people who have

9:46

been victims of revenge porn and what have

9:48

you. And then the third here also is

9:50

politicians. Everybody's screaming from

9:52

the rooftops about that Biden

9:54

deep fake robocall telling people not to

9:56

vote. And I think that in the election year, we're gonna

9:58

be seeing a lot more. play on that.

10:01

And we haven't only seen

10:03

calls for legislation here, we've

10:05

seen actual legislation being proposed,

10:07

several different bills, federally that

10:09

on their surface seem to

10:11

be trying to do a

10:13

good thing. But unfortunately, it seems

10:16

would potentially make the situation

10:18

worse or make the overall

10:20

status of artists in particular of the

10:22

next generation of Taylor Swift worse than

10:24

it is today. First of

10:27

all, can you tell me more about the Biden

10:29

deep fake call? Yeah, so

10:32

in New Hampshire, I think

10:34

a couple of weeks back

10:36

during the primary, there was

10:38

a really interesting operation that

10:41

was carried out, I think it was

10:43

first flagged by an election

10:45

organizer whose phone number was

10:47

spoofed to call people in

10:49

New Hampshire and play

10:51

a recording of what

10:54

sounded like Joe Biden telling them not to

10:56

turn out and vote. And

10:59

the election worker found election organizer found out

11:01

that this had happened when people started calling

11:03

her number back. And,

11:06

and that

11:09

is very obviously a concerning

11:12

thing when you can take the voice of

11:14

a politician and tell those

11:16

who support them to do something that they

11:18

don't want want want them to do.

11:20

And I'll also add to

11:23

that simultaneously, we also are now seeing

11:25

politicians like Donald Trump claiming that photos

11:27

they don't like that don't make that

11:30

don't make him look good are alternative

11:33

AI deep fakes and that those photos

11:35

aren't real. So it's just it's a

11:37

mess is what it is. We're

11:40

entering into a lovely world where I mean,

11:42

not only can you not trust audio

11:45

video that you see, especially online,

11:48

but then people can use it as a

11:50

way to dismiss any information they don't like.

11:53

I was actually thinking about what is

11:55

a term for that last phenomenon you

11:57

just described because in I

12:00

know in activist spaces, there's a term

12:02

called fed jacketing or snitch jacketing, which

12:04

essentially means when you when people start

12:07

accusing one another of being secret informants

12:09

or secret like government agents.

12:12

And so I feel like in the same way we're

12:14

about to because of the fact that the

12:17

media environment is so

12:19

completely, like chaotic and

12:21

untrustworthy. And because of AI playing a

12:24

huge part in that it feels like

12:26

we're about to start seeing like AI

12:28

jacketing or some some I don't

12:30

know, I was trying to come up with like a

12:32

better term for it. But like, I've been noticing this

12:34

recently. And also with your reporting,

12:37

Matt on pow world, there were

12:39

people that pow world is this is

12:42

this video game that we've been talking about

12:44

where it's essentially Pokemon, but like it's made

12:46

by this other company, and they kind of

12:49

sort of kind of look like Pokemon, but

12:51

they're not. And people hate this game, because

12:53

it's like like jolting off of Nintendo's

12:55

IP. And some people were accusing it of

12:57

using AI generated assets, but there was no

13:00

evidence for that. So it's just kind of

13:02

like another example of people, you

13:04

know, using this like environment

13:06

of distrust to like, pro

13:09

accusations. The power world

13:11

one is a really low stakes good example.

13:14

Because I remember watching

13:17

that, it was one account

13:19

on Twitter accused of

13:21

the game of stealing

13:23

assets and using generative AI to change

13:25

them just enough, and then

13:28

provided like screenshots that

13:30

proved quote unquote, that this is what this is

13:32

what had happened. And it's

13:35

one of those things where like you hear about it,

13:37

and then it becomes just cut part of the conversation

13:39

around the game. Like, oh, they used AI, did you

13:41

know that they used AI isn't this terrible, etc, etc.

13:44

And the company had previously published a

13:46

game that used generative AI in it,

13:48

it was like a social deduction game

13:50

where you're trying to figure out what

13:53

portrait was created by

13:56

generative AI, I'm kind of like

13:58

an among us using midget. journey,

14:00

basically. Then the person that

14:02

had done the original tweet citing the evidence admitted

14:04

that they just made the whole thing up later,

14:07

and that it was complete bullshit. So

14:09

yeah, I think AI jacketing is a good term. They

14:12

were also just posting screenshots of

14:15

the CEO's tweets saying like, oh

14:17

yeah, I think the generative AI

14:20

is interesting. That was

14:22

a hard set. And then we

14:24

also see, I've seen because I'm also

14:26

an author and a huge publishing nerd,

14:28

we've seen cover authors

14:30

and publishing be accused

14:32

by the cancel

14:35

mob that exists within publishing,

14:38

rightly sometimes and also wrongly sometimes,

14:40

of using AI to

14:43

make book covers when, in reality, this

14:45

is an artist who's worked a very

14:47

long time to make something extraordinarily beautiful

14:49

and that sort of controversy

14:51

and just the sheer mental

14:55

and emotional weight of that sort

14:57

of attack. They're really serious and

14:59

pretty harmful, this whole idea

15:01

that if you touch AI at all, then

15:03

we should throw you off a cliff. I

15:06

was following one of those stories last year

15:09

and didn't get a chance to really report

15:11

out on it. There

15:13

was mostly happening in like Facebook groups. And

15:16

it was a freelance artist who

15:19

they people had decided had used AI

15:21

to generate her covers. And

15:23

I could not tell. I

15:25

couldn't quite decide

15:28

to my own satisfaction

15:30

if she was using

15:32

generative AI or if

15:35

she was just a very talented artist. There

15:37

weren't a lot, some of the tells were

15:39

there. Maybe she used it to

15:41

do a pass. But it was

15:43

difficult. It was really hard. And that's ultimately

15:45

part of why I didn't report on it.

15:47

Because I couldn't make a determination either way.

15:49

And there have been

15:53

scandals around the use of generative AI

15:56

by major publishers in the past few

15:58

years. Like

16:00

this stuff does happen. Everyone's

16:02

getting paranoid. It's terrible. I

16:06

think that that is symptomatic

16:09

to a certain extent of the fact that we are

16:11

grappling with this thing as a culture and how it's

16:13

going to be and how we're going

16:15

to think about it and what are the acceptable

16:17

ways to use it and what are the

16:20

acceptable ways to criticize it. And

16:23

we're not there yet collectively. That

16:26

growing pain to me, well, I think very

16:28

painful for a lot of people. It seems

16:30

relatively normal in terms of technology.

16:34

We do kind of go through this every time there's a sea

16:36

change, right? And it's going to take I

16:39

think on the legislative side, unfortunately, probably years to

16:41

catch up and we're probably going to do some

16:44

things that we shouldn't do. Yeah.

16:48

Janice, you were going to say something about paranoia.

16:50

Oh, just that, you know, the

16:52

kind of paranoia is made worse by the fact

16:54

that, you know, it actually

16:56

is happening. Like, people are

17:00

generating, like using generative AI to

17:02

do all these kind of like deceptive things. And

17:04

so, you know, it's

17:07

because of that that the sort

17:09

of accusations become effective.

17:11

Like, it wouldn't be effective if we

17:13

weren't currently living in a

17:15

time when it was so trivially

17:18

easy to generate visual propaganda using

17:21

a system created by

17:24

large corporations that were built using

17:26

millions of other images taken without

17:28

permission. And playing

17:32

off of that paranoid surface area, one

17:34

of the most common quote-unquote solutions that

17:36

I see proposed to try to

17:39

make the internet

17:41

trustworthy again or some sort

17:44

of Don Quixote type

17:46

task is to enforce

17:52

labeling machines across every

17:54

tool that uses generative

17:58

AI to make images or

18:00

video or music or what have you. And on

18:03

the surface that could seem like a

18:05

good idea but we have a lot

18:07

of concerns about the reality that in

18:09

many places generative AI tools will be

18:11

made that don't have those AI

18:14

labeling routines and

18:16

that if we

18:18

come to trust that what is labeled as AI

18:20

is AI and what

18:22

is not labeled as AI is

18:24

real that the

18:27

ground for the most sophisticated and

18:29

most manipulative actors with the highest

18:31

stakes to be able to fool

18:34

us. And there's all

18:36

these secondary cascading effects

18:38

of content filtering on platforms

18:40

and whether or not we're surveilling

18:42

and collecting data on who's making

18:44

AI images at the point of

18:47

creation and there's just there's

18:49

a lot a lot there and I see a

18:51

lot of folks trying to go down that path

18:53

and I think

18:55

like you said Matthew it's really

18:58

important to think these through all

19:01

the way because

19:05

even a little bit of life we should label

19:07

it could have a lot of unintentional effects. What

19:10

would that even look like? Are you

19:12

talking like a watermark? What's been suggested?

19:14

Yeah they're talking about

19:17

visual watermarks they're

19:19

talking about invisible

19:22

cryptographic watermarks that then your

19:24

Instagram or what have you

19:26

could could read and then pull

19:28

a label in the app from that

19:31

but the reality of

19:33

the internet to get around

19:35

restrictions such as these is

19:37

well established. I

19:41

want to go back a little bit and talk about more in-depth

19:43

some of the the legislation that's being proposed

19:45

proposed you know some of it's being proposed

19:48

at the state level it's being proposed at

19:50

a federal level I think

19:52

the most well-known one is probably the no

19:54

fakes act can you walk us through some

19:56

of this stuff and is

19:58

it good is it bad I

20:01

think it's, I think that it is coming

20:03

from the right place with legislators. So

20:05

the No Fakes Act, I think is a good one to

20:08

walk through. There's

20:11

also the No AI Fraud Act, which

20:13

is maybe the next evolution of No

20:15

Fakes, or at least

20:17

proposed more recently. And

20:19

some of the terms in there are a little bit more reasonable.

20:22

But this approach that both

20:24

of those acts in particular are taking right

20:26

now is the idea that

20:28

they should create a new federal

20:31

intellectual property right to

20:34

everyone's likeness. So

20:36

that's to your face, to your

20:38

voice, your

20:41

mannerisms, essentially a right to your body.

20:45

And it's representation in the

20:47

digital world. And

20:50

the idea behind doing that is that if you have a

20:53

right to that, and it's your intellectual

20:55

property, then you can sue people who

20:57

use it without your permission. But

21:01

the problem in creating that new right, and

21:03

with this legislation, the first problem that I

21:05

have with it is that it's

21:08

establishing a new transferable intellectual

21:10

property right. And

21:14

I believe that it's incredibly important for

21:16

us to really question whether

21:18

we want to let people sell

21:20

the rights to this representation of

21:22

their physical body in

21:24

the digital realm and video and

21:26

music, what have you. And

21:29

it's funny because it felt like some

21:32

sort of Swiss Deep Groundhog Day when

21:34

the deep fakes were of Taylor Swift,

21:37

because I had just been walking

21:39

through the impact of the OAI

21:44

fraud act in particular, in the

21:46

context of Taylor Swift's own experience

21:48

as an artist, where

21:50

early on, she signed away

21:52

that the rights, that intellectual property rights

21:55

to her master recordings, which

21:57

were ultimately acquired by the Sky Scooter Braun,

21:59

who's You know my very much for

22:01

reasons that you can google. At an

22:04

end to think about it at that

22:06

time there at also Van or right

22:08

to Taylor's face. Or a right to

22:10

the sound of her voice. And.

22:12

The incentive Us Mater Content companies are are

22:15

the rich men of Hollywood's You get as

22:17

many rights as a chance that they can

22:19

exploit them maximally would have probably meant of.

22:22

that contract that she signed at that time

22:24

would have included her face in her voice

22:26

in The Scooter Braun could essentially on those

22:28

two to the. And

22:31

that that is incredibly concerning and and

22:33

to me is. Is the fundamental

22:35

no holds barred as long

22:37

as ridiculous as. Fact:

22:40

I hit the because What it what it does is

22:42

it sets off. All these artists that it's

22:44

trying to protect and said to

22:46

be exploited without their consent. Ah

22:48

by a eyes along and that

22:50

a i it's being wheeled that

22:52

by the person who has acquired

22:54

the rights to their body. In.

22:57

Also, wouldn't that mean like you would

22:59

have to register yourself in a government

23:01

database of some kind? Before.

23:03

You were able to pursue a noise

23:05

a criminal case against some of us

23:07

making within a form of you. Eat.

23:10

You know I don't They did. I think

23:12

that's because it's. An. Intellectual Property

23:14

Rights that it actually was the

23:16

something that you have automatically but

23:19

you are correct their that the

23:21

rights is the right to see.

23:24

It's. Not the right the has

23:27

the eyes aren't as you

23:29

removed from the internet. It's

23:31

not a take down is

23:33

the right to get into

23:35

a legal battle with various

23:37

entities. Some of them are

23:39

more than others. Under. The. Hindus

23:44

and protracted. Legal. Your

23:48

switches Mosquirix, Have to.

23:50

Seriously. Demons. And for human being.

23:53

How do you think we should legislate of we

23:55

handle this? i don't

23:57

have all the answers here for sure

24:00

incredibly complicated, but one

24:02

of the places to start is with

24:04

the existing laws that a bunch of

24:06

these people have been affected that we're

24:08

already talking about are already suing under.

24:11

Most states have a right

24:13

of publicity law that allows them to

24:15

sue, or that allows you know, celebrities

24:18

or public figures to sue if people

24:20

misuse their images in some sort of

24:23

manner that is commercial or could be construed

24:25

as commercial. And then at the same time

24:27

we have defamation law

24:29

which can often be a

24:32

way to, for average people,

24:34

to sue those who

24:36

humiliate them and, well,

24:39

it protects celebrities less. So between

24:41

those two, depending on where you

24:43

live and in the

24:45

majority of states, there's some good stuff there.

24:47

There's some good mechanisms, at least in terms

24:49

of if you want to get a lawyer

24:51

and if you want to sue the person

24:54

who's causing a misery. But

24:57

still that doesn't actually give victims

25:02

of these deep fakes a

25:04

real-time way to say, hey, wait,

25:06

get this disgusting part of me off

25:09

the internet, this is humiliating me and

25:11

it's running everywhere. And

25:14

while it pained me, because I

25:16

know how extensively something like the

25:19

Digital Millennium Copyright Act has been

25:21

abused, they sort of noticed and

25:23

take down systems. I'm

25:25

talking to a lot of people and none of us really

25:27

see another way

25:29

to make something that

25:33

is actually responsive to the harms that people are

25:35

going to be experiencing. But what we can get

25:37

right this time, and I think with

25:39

Donald Trump right now claiming that actual

25:41

photos of him are deep fakes, is

25:43

that unlike the DMCA, we can build

25:45

a system where there are actual consequences

25:47

if you abuse it. If you

25:50

say that it's an embarrassing video, if you were a

25:52

video of police misconduct or what have you

25:54

is a deep fake and it actually isn't.

26:00

Oh, that's interesting. What do you I

26:03

didn't even think about the possibility

26:05

of a public person getting

26:07

into trouble for lying about a real image being a

26:09

deep fake. Oh, yeah, that's coming.

26:13

What has anyone proposed legislation or you

26:15

ordered this just like conversations? This

26:18

is conversations because we

26:20

I think that a lot

26:23

of the people who are looking at

26:25

no fakes or no AI fraud or

26:27

what have you and and saying that

26:30

these are these are terrible laws with

26:32

unintended extremely harmful consequences are also really

26:34

feeling for the reality that people are

26:36

going to face with these technologies and

26:38

that and knowing that we need to

26:40

do something. Happily,

26:43

we're having much more of a proactive

26:45

conversation here amongst ourselves about what we

26:47

do want. Then, then I think we

26:50

have maybe it was previous online revolutions.

26:53

That was one of the things that struck me

26:55

initially I wrote about the no fakes

26:57

and then there was another law in Tennessee that

26:59

I think was proposed the same day, which

27:01

is that, you know, and this is like what I wrote

27:04

about in the article I wrote a couple of weeks

27:06

ago, which is like, it just, it seems like

27:08

the just of these laws is sort of intended

27:11

to protect celebrities, and

27:13

maybe the rest of us to kind

27:15

of sort of, and

27:18

that's kind of like where, where I

27:20

came out this from which is like the fact that

27:22

like, you know, a lot of people have been

27:24

talking about this as a concern for a while. But

27:27

now that Taylor Swift is mad. Now

27:29

that like the weekend and Drake are

27:32

mad about this, and it's not just

27:34

you know, it's not

27:36

just like photos. It's also

27:38

like music and, you know, voice rights

27:40

music, and then all other kinds of

27:42

stuff that could be considered you know, intellectual

27:45

property or, you know, personalized

27:47

sort of like intimate representations

27:49

of someone's person. It

27:52

seems to me like that's

27:54

where this always starts and ends

27:56

when we get quote

27:58

unquote privacy protection. or

28:00

something that is supposed to at least in

28:02

theory be like protecting privacy

28:05

is that it's generally winds up

28:07

protecting Famous and rich

28:09

people and not doesn't do a whole

28:11

lot for regular people

28:13

who are facing abuse and harassment

28:15

and you know sexual violence

28:17

and You know the

28:20

copyright system as you were just mentioning like

28:22

the copyright system you're saying like oh You

28:24

know we keep looking at this like we

28:26

don't see any other way of enforcing that

28:28

Like what what would be different about this?

28:31

Compared to like you know the fact

28:33

that you know when people when

28:36

artists for example have people Profiting

28:38

from their art that

28:40

they release like music or otherwise, you

28:43

know, there is a motive for dress But it's not

28:45

very accessible unless you have a lot of money to

28:47

lit litigate it So what's the fifth here when it

28:49

comes to this stuff if in the past this has

28:51

been kind of a status quo That's

28:54

a great question Because we are

28:56

really swimming upstream against the

28:59

headline grabbers here if

29:01

something horrible is being done to Taylor Swift and

29:04

By God Josh Holly can trot out there with a

29:06

bill and and and wave it in the air and

29:08

it's gonna get covered You know

29:10

all over creation The motive

29:12

the motive there is really clear

29:15

and straightforward politicians like flaws that

29:17

grab headlines they like flashy partnerships

29:19

with celebrities and and and I

29:22

would also say that the the

29:24

lobby of those IP

29:26

rights holders the major labels and

29:29

Publishers and content companies and what have you

29:31

is extremely powerful and really well organized and

29:33

from the moment that this all blew up

29:36

They've been in legislative offices, you know gunning

29:38

for a bill that's gonna benefit, you know

29:40

the Universal Music Groups of

29:43

the world and yeah, and that's why For

29:47

myself I turned towards and I think that there

29:49

are also legislators who who are thinking

29:51

in this way better

29:54

tools for every

29:57

everyday people because I

29:59

don't think that there's going to

30:01

be an effective way to censor, or

30:06

we can't put the router back

30:08

in the hat with AI. What

30:10

we need is proactive tools to address

30:12

it in a way that is minimally

30:15

invasive when it comes to surveillance

30:17

and censoring speech and slapping upload

30:19

filters across the whole internet isn't

30:21

the right thing either. I would

30:23

look at something like, well, we've

30:25

got Google reverse image search and

30:27

we know that that works pretty

30:29

good. And we've got the DMCA

30:31

and that, there's

30:35

an established protocol for that. And we've

30:37

got this idea that I've learned a lot

30:40

since that legislation went in.

30:42

And so can we slap something together

30:45

that doesn't reinvent the wheel and

30:47

just gives people the right to say, hey, this is

30:50

a horrific photo, a

30:52

horrific fake photo of me. Please

30:55

grab it off the internet and

30:57

I can make that request in a

30:59

way that the platforms have to

31:01

be accountable to. And that's the other thing, it's

31:03

really hard to be heard as an individual

31:05

when platforms are dealing with so many

31:07

users. About that

31:09

point on platforms being accountable, this is kind

31:11

of like something I say a lot in

31:13

when I'm talking about this topic is that

31:15

we're kind of addressing a symptom of

31:18

a problem here and not the actual problem.

31:20

And the problem is that we have all

31:23

these giant tech companies that

31:25

are producing this technology and they basically

31:27

have no regulation and they're kind of

31:29

just doing whatever they want. And

31:33

that's, there's even like these

31:35

like lobbying groups like Elon Musk has

31:37

like this like AI Institute that's like essentially

31:39

saying, you must let us develop AI and

31:41

if you don't, then you're killing people, people

31:44

will die. And so, that's kind of

31:47

like what I always frame this around is

31:51

like we're dealing with a symptom and

31:53

not the like actual problem, which is that

31:56

like, and even the

31:58

way in which these tech companies address. this

32:01

problem sometimes is very much like band-aid

32:03

oriented. Like I was writing an article

32:05

a couple of weeks ago about the

32:09

filtering system on some of these

32:11

things like on OpenAI, like

32:13

has been constantly needing to patch

32:16

chat GPT and like

32:18

Dolly and all these like image generators

32:20

because people keep finding ways

32:23

to like get

32:25

past the like content filter

32:27

system that like prevents them

32:29

from generating certain types of

32:32

images through all these kind of tricky ways

32:34

and it's just this cat and mouse game.

32:37

And you know, when it

32:39

comes to some of these, I was reading this paper

32:41

and when it comes to some of these systems what

32:43

they're actually doing is that they're just, they're

32:46

still generating the content and then they're just not

32:48

showing it. So it's not even that

32:50

they're like preventing the content

32:52

from being created in the first place, they're just filtering

32:55

it out. Like they're just saying, you

32:57

know, the equivalent of like, I

32:59

can't do that Dave is

33:01

what the policy is right now is what the

33:03

sort of mitigation policy is for a lot of

33:05

these things. But really it's off

33:07

camera sketching that image you asked for and

33:09

just storing it in a digital warehouse somewhere

33:12

and all the forbidden images you'll never

33:14

be able to see. Yeah,

33:16

but it's like, even if nobody sees that

33:19

image that's indicative of a larger problem, which

33:21

is that like, we don't actually know how

33:24

to stop that from happening because the

33:26

training has already occurred.

33:28

These systems are already built on

33:31

top of billions of images that

33:33

were taken without permission. And

33:35

you know, some of them, we wrote another story about

33:37

this a couple of weeks ago, LAION,

33:40

which is a probably one of the

33:43

most commonly used image databases used in

33:45

generative AI. It

33:47

contains billions of images that are taken from

33:49

web scraping. And it was found that I

33:51

think about 3000 instances of CSAM, of

33:56

child exploitation were found in the

33:58

state in this massive. database and

34:00

it's like you know that's what

34:04

we're dealing with here it's like the

34:07

sort of like base problem has already

34:09

occurred where we're just getting the results of

34:11

it now and you can filter the results

34:13

but that doesn't ultimately like solve the fact

34:16

that like where this all came from. All

34:20

right so I'm just going to pause there for a break

34:22

we'll be right back after this. Are

34:28

you ready to enhance your future in tech?

34:30

Then it's time to make your move to

34:33

the UK the nation that

34:35

has more tech unicorns than France

34:37

Germany and Sweden combined the

34:40

nation that was third in the world

34:42

to have a one trillion dollar tech

34:44

sex evaluation the nation

34:46

where great talent comes together

34:49

visit gov dot uk/great talent to

34:51

see how you can work live

34:54

and move to the UK. We're

35:00

driven by the search for better but when it comes

35:02

to hiring the best way to search for a candidate

35:04

isn't a search at all. Don't search

35:07

match with Indeed. If

35:09

you need to hire you need Indeed. Indeed

35:12

is your matching and hiring platform with over

35:15

350 million global monthly visitors according to

35:17

Indeed data and a matching engine that

35:19

helps you find quality candidates fast.

35:22

It's the busy work. Use Indeed

35:25

for scheduling, screening and messaging so

35:27

you can connect with candidates faster

35:29

and Indeed doesn't just help you

35:31

hire faster. 93% of

35:33

employers agree that Indeed delivers the highest

35:35

quality matches compared to other job sites

35:38

according to a recent Indeed survey. When

35:40

I was looking at Harrison then it was so

35:42

slim and overwhelming I wish I had used

35:44

Indeed. Leveraging over 140 million

35:47

qualifications and preferences every day, Indeed's matching

35:49

engine is constantly learning from your preferences.

35:51

So the more you use Indeed the

35:53

better it gets. Join more than 3.5

35:56

million businesses worldwide that use Indeed to

35:58

hire great talent. And

36:01

listeners of this show will get a $75 sponsored

36:04

job credit to get your jobs and more

36:06

visibility at indeed.com Cyber

36:09

just go to indeed.com/cyber right now and

36:12

support our show by saying you

36:14

heard about indeed on this podcast Indeed

36:16

comm slash cyber terms and conditions apply

36:19

Need to hire you need indeed.

36:21

Okay, there's cyber listeners Matthew here. This

36:23

episode was brought to you by delete me

36:27

Few years ago. I did some reporting that Made

36:29

quite a few people pretty angry I got

36:31

a few death threats and I started to worry about

36:34

how much my private information was available in public so

36:36

we've got a safety team here advice and that

36:39

helps journalists navigate these kinds of situations and The

36:42

single best thing they did was sign me up

36:44

for delete me So, you

36:46

know, how would you like to wake up one day and

36:48

discover your bank account has been emptied or

36:50

get an overdue notices For credit cards you never applied

36:52

for or worse You

36:54

don't realize how much of your information is

36:57

available to scammers on the internet and how

36:59

susceptible you are and your family is

37:01

to identity theft and fraud Now

37:04

you can get your data removed with delete

37:06

me These are just two of

37:08

the many reasons that I personally recommend delete

37:10

me delete me is a subscription service that

37:12

removes your personal info from the largest people

37:15

search databases on the web and

37:18

in the process Helps prevent potential ID theft

37:20

doxxing and phishing scams and this is not

37:22

just a one-time thing either It is always

37:24

working for you. I've been

37:26

using it for a few months now

37:29

and I'm constantly getting updates about

37:31

where my information is popping up

37:34

how to remove it and One of the

37:36

best parts a lot of the stuff just gets

37:39

automatically removed for me to put it

37:41

simply delete me does all the hard Work of wiping

37:43

you and your family's personal info off

37:45

of the web data brokers hate delete me

37:47

when you sign up for delete me It

37:49

immediately goes to work scrubbing your all your

37:52

personal information from data broker platforms When

37:55

you sign up delete me immediately goes to

37:57

work scrubbing all your personal information from data

37:59

broker Your personal

38:02

profile is no longer there to sell. So

38:05

take control of your data and keep your

38:07

private life private by signing up for Delete Me. Now

38:10

at a special discount for our listeners.

38:13

Today get 20% off

38:15

your Delete Me plan

38:17

when you go to

38:19

joindeleteme.com/cyber and use promo

38:21

code cyber at checkout.

38:23

The only way to

38:25

get 20% off is

38:27

to go to joindeleteme.com/cyber

38:29

and enter code cyber

38:31

at checkout. That's joindeleteme.com/cyber

38:33

code cyber. AI

38:36

might be the most important new computer

38:38

technology ever. It's storming every industry and

38:40

literally billions of dollars are being invested.

38:43

So buckle up. The problem is

38:45

that AI needs a lot of speed and processing

38:47

power so how do you compete without costs spiraling

38:49

out of control? It's time to

38:51

upgrade to the next generation of the cloud, Oracle

38:54

Cloud Infrastructure or OCI. OCI

38:57

is a single platform for your

38:59

infrastructure, database, application development and AI

39:02

needs. OCI has four

39:04

to eight times the bandwidth of other

39:06

clouds, offers one consistent price instead of

39:08

variable regional pricing and of course nobody

39:11

does data better than Oracle. So

39:13

now you can train your AI models at twice the

39:15

speed and less than half the cost of other clouds.

39:18

If you want to do more and spend less like Uber,

39:21

8x8 and Databricks

39:23

Mosaic, take a

39:25

free test drive

39:27

of OCI at

39:29

oracle.com/cyber. That's oracle.com/cyber.

39:31

oracle.com/cyber. Alright

39:35

cyber listeners, welcome back. We are

39:37

back on talking about legislation and

39:39

artificial intelligence. On

39:42

CSAM, I noted that several of the

39:45

local legislation was

39:48

pretty narrow. I think

39:50

Ohio specifically is

39:53

just about miners and

39:56

I don't know why they like zeroed

39:58

in on that specific thing and didn't

40:00

do that. something for adults, but the

40:02

proposed like Ohio draft legislation would narrowly

40:05

target sexual

40:08

like basically deep fakes made of people under

40:10

the age of 18. I thought

40:13

that was an interesting carve

40:15

out, no protection for anyone above the age

40:18

of 18. Legislators do

40:20

often find it easier to

40:22

pass bills that are about

40:24

saving the children. That's a

40:26

common technique that we've experienced

40:28

across the range of the issue

40:30

areas that we work on. At

40:34

the same time, I haven't reviewed

40:36

that bill specifically and what

40:39

its enforcement mechanisms are, but

40:41

I do think that narrowly

40:43

targeted legislation that is

40:46

focusing on specific harms is not a

40:48

terrible direction to go generally.

40:50

You just have to make sure that you

40:52

aren't burning down the whole internet

40:55

or important resources from our religious communities

40:57

or our online speech

40:59

or what have you in pursuit

41:03

of those protections. I

41:08

feel like we're going to. I

41:10

feel like it may not be Taylor

41:13

Swift, but something is going to happen

41:17

to somebody that has power that they are

41:19

not going to like. Then

41:22

there will be a law that oversteps.

41:25

It feels like what is going to happen

41:27

in the next few years. Am

41:30

I completely out of bounds? Oh,

41:33

I'm been waiting for that never since this

41:35

whole hell of blue. I

41:37

started up over a year ago now. Absolutely.

41:41

We're going to see something massively

41:45

impactful to somebody. I mean, who's

41:47

more powerful than Taylor Swift? I

41:49

don't really know. But happening to

41:51

a legislator or with

41:55

a compelling impact on democracy or

41:57

what have you. That may force us

41:59

to. to confront faster than

42:01

we're ready what we need

42:03

to do to mitigate

42:06

these harms. And

42:08

I think that these early

42:11

legislative drafts and the criticisms of them and

42:13

what have you are

42:17

getting civil society

42:21

and forcing legislators into conversations where

42:23

we are talking about how

42:26

not to throw the baby out with the

42:28

bathwater and that part of

42:30

the reason that we are where

42:32

we are today with the mass

42:35

image scraping and the biometric

42:37

databases and just being able

42:39

to log in on some

42:41

data broker website. And

42:45

I was thinking of a case

42:47

where a family was almost tricked

42:51

into sending mail money for their

42:53

son after somebody got

42:55

all the phone numbers for the

42:57

family and impersonated the son's voice

43:00

and this, that and the other thing. A lot of

43:03

the vulnerabilities we're facing here is just because we don't

43:05

have data privacy and the

43:07

US is a sort of substantial way. We

43:10

really haven't grappled with one of

43:12

the worst and most extreme harms

43:15

of the era of

43:17

the internet that some say that we're

43:19

leaving. And

43:21

I would say too, that you coming

43:23

into the election thinking about it can

43:25

return a litica but

43:28

with this technology and the ability to

43:30

analyze and iterate at scale is extremely

43:33

concerning. And again, that

43:35

goes back to privacy and

43:38

that we don't have it and then that data

43:40

is just laying there to be exploited. Yeah,

43:44

that's a good point that we've kind of, we've moved on

43:46

to like a new phase of the internet having

43:49

not solved any of the problems of

43:51

the previous phase of the internet. And

43:54

now we have, and I always, I keep

43:56

saying that like the sort of deep fake phenomenon

43:58

is gonna be like the... death of

44:00

not the literal death, but it's

44:02

going to be the end of like boomers

44:05

like falling finding like Taylor Swift

44:07

videos like saying here click here

44:09

for this cook for a free

44:12

cookware set which actually happened. That

44:15

was another story we did a couple weeks ago

44:17

about somebody deep thinking

44:19

Taylor Swift's voice and like getting

44:21

people to like click on a

44:23

link for you know cookware set

44:25

that wasn't being offered and didn't

44:27

exist and it was like you

44:29

know it's so trivially easy. That's

44:31

the stuff like the sort of industrial scale

44:33

that comes into play when which

44:36

is all built off of the fact that we didn't

44:38

really address a ton of these

44:40

issues before they became like able to be

44:42

automated. Oh I would say that

44:44

it's the death no for for Millennials

44:47

too. Oh yeah.

44:50

I think the the the the Millennials

44:52

are the ones being targeted by Taylor Swift. The

44:54

boomers are all getting phone calls. That's

44:57

true. So this is just like

44:59

this is the advanced robocall for the people

45:02

that grew up online and I wonder I

45:07

wonder how much of all of this will burn down

45:10

to avoid getting hurt and that's

45:12

one of my big concerns. So I

45:15

also don't live but also I don't know I don't know what you

45:17

do. I just I

45:19

simply don't. Well

45:21

I would say that there are a fair amount

45:23

of parallels. I remember back

45:26

in high school whether or not you're allowed to

45:28

cite Wikipedia and whether you can trust the Internet

45:30

whether you can trust what you read on the

45:32

Internet and and that sort

45:34

of neat media literacy and the

45:37

the existential dread of a website

45:39

that you can change instead of a newspaper that you

45:41

can't was really present

45:43

in that time. And this

45:47

to an extent feels like

45:49

that although we're talking about

45:51

much more convincing mediums to

45:53

our little animal brains that

45:55

have not evolved for any

45:58

of this. And

46:01

so while I'm not sitting

46:04

in the most optimistic place here, I

46:06

also don't feel that it's completely foreign

46:08

territory and to me at least is

46:10

encouraging. It's a

46:12

fair point. We've been through big,

46:15

frightening changes about the

46:17

way the information sphere works before and though

46:19

to Janice's point we may have not solved

46:22

some of the previous issues we are surviving

46:24

and still using the internet. The

46:29

ways we use the internet I think will definitely change. And

46:32

that's my transition into

46:34

talking about the

46:37

back half of the conversation today. Just

46:40

kind of how we started talking about

46:42

this offline Janice, which is that the

46:44

AI thing is interesting because

46:47

it's in a boom cycle right

46:49

now that I think

46:51

looks very different to you and I because

46:53

we're journalists and we have our

46:55

inboxes are very strange. It

46:58

looks a lot on the surface like

47:01

what Web3 and crypto looked like a

47:03

year ago, right? Could

47:06

you tell me about your

47:09

thoughts on how AI

47:11

has become the new Web3? Oh,

47:14

for sure. I mean, I guess the

47:16

best way of phrasing this is like

47:19

I invite, well, okay, I don't invite

47:22

other people to look in my inbox, but like

47:25

if I were to give you a

47:27

sampling of my email inbox on

47:29

a weekly basis, you would

47:31

see countless pitches, some

47:33

of the worst pitches that I've ever seen in

47:36

my life from all these companies that didn't exist

47:38

six months ago. And

47:40

oftentimes, and I've been meaning to do a

47:42

more sort of like in-depth

47:44

investigation about this and maybe even write about

47:46

it. A lot of times I

47:48

will look up these companies that didn't exist six months

47:51

ago and I will find that their founders are

47:53

the same people who did crypto

47:55

startups and Web3 startups like a

47:57

year ago. And, you know,

47:59

the… this kind of driving home for me,

48:01

the fact that we are

48:03

in this kind of looping hype

48:06

cycle when it comes to tech and when it

48:08

comes to the sort of like

48:10

economics and like the phase

48:12

of capitalism that we're in where

48:14

there's really no new ideas. Everyone

48:17

is just kind of like tweaking the ideas

48:19

a little bit each time and trying to

48:21

sell their new grift. And

48:23

you know, AI is the new grift. There is

48:25

a difference I should mention. I

48:28

think that unlike NFTs, for

48:30

example, there is an actual use

48:32

case in a lot of cases for AI

48:35

to automate things. I think that like the

48:38

allure of automation is pretty undeniable.

48:40

And I think that especially

48:43

when it comes to companies

48:46

that are looking for ways to cut

48:49

costs and do like the sort of

48:51

siren call of that is just completely

48:54

unable to be resisted. Beyond

48:57

that, there are actual

48:59

use cases in which AI can be,

49:01

I think, used to

49:04

help people. However,

49:07

you know, it has

49:09

all these problems that we've been talking about. That's

49:11

now and in a lot of ways,

49:14

the sort of like cycle that

49:17

we've gotten into where it seems

49:19

like it's becoming faster. Like Mark Zuckerberg was

49:21

pitching everybody on the metaverse like a year

49:24

ago. Was it

49:26

that recently and

49:28

already divesting from that? And it's, you

49:31

know, it's kind of like a weird

49:33

death spiral type vibe right now in

49:35

tech. Yeah, he's

49:37

got to shed what, 22 percent of his workforce to

49:39

get to get returns back

49:41

up on the company, which

49:43

worked, you know, short

49:46

term actions produce lovely short term results.

49:48

Yeah, everybody's in short term right now.

49:51

That's the thing. Everybody is like it's just

49:53

what's the what's the latest buzzword soup that

49:55

we can sell people on? I

49:57

think everyone's in short term most of the time. It's

50:00

kind of the way the business operates. It's

50:03

what it feels like to me anyway. The

50:05

way business operates. Yeah, I recall

50:07

the moment when I was reading

50:09

the crypto VC is

50:11

moving to AI article and business

50:16

and insider who knows what it was. And

50:20

my first reaction was just, Oh,

50:22

here, here we go. Here we go. And

50:25

the new AI

50:27

thought leadership, human

50:30

rights, whatever

50:33

human artistry campaign, but it's

50:35

the RIAA and these

50:37

sorts of like

50:40

new organizations that

50:42

are being being set up often by like

50:44

VC capital or established and to use their

50:46

what have you to sort of ask to

50:48

a turf on the work that the

50:51

activists have been doing for years.

50:54

It was from our perspective,

50:57

it was really profound over

50:59

the past year to see

51:02

how erased

51:04

so much of the like decades of

51:06

work of folks with the Algorithmic Justice

51:08

League or like our work

51:10

on facial recognition was

51:13

from the larger conversation.

51:15

And the gut instinct there

51:17

was 110% to build something

51:19

new rather than

51:21

to go to the analysis and

51:24

advocacy that had been happening for years

51:27

upon upon years and

51:29

this faith and actually a lot of

51:31

good, good, good friends and good starting

51:33

points for ways to ways to think

51:36

about this and are going to write

51:38

for it. It would actually help AI

51:40

be that force for good instead of

51:42

just another tool to Sophia.

51:46

It's kind of Silicon Valley's attitude all over, right?

51:49

It's trapped in this eternal present moment

51:52

with no sense of the

51:55

history that it's attempting to disrupt,

51:57

quote unquote, right? It

51:59

doesn't. understand that whatever a lot of

52:02

the products and services it's trying to

52:04

offer are already being offered in a

52:07

way that the marketplace deems efficient and

52:10

that you know if

52:12

you try to release XY or Z product you're just

52:14

going to end up where the market

52:16

is now but in 30 years

52:18

time and after hurting a lot of

52:20

people and maybe not making nearly as much money as

52:22

you thought you would. And then

52:25

they are it's like it's capitalism

52:27

again they're there and they're devised to

52:29

do that that is that is how

52:31

they are supposed to operate legally out

52:34

of obligation to their shareholders which is

52:36

which hurts you know. I just

52:39

saw somebody a good example of what we're

52:42

talking about I just saw somebody a screenshot

52:45

a tweet of someone complaining because

52:47

Revel the like startup that has I don't

52:49

know if they have this where you are

52:52

Leah but it's like the sort of like

52:54

mopeds that you can rent with an app

52:57

and drive around Revel is apparently

52:59

closing in New York City so they're

53:01

not gonna they're they're done and

53:04

somebody was complaining about this thing great

53:06

now that Revel isn't around there's

53:09

no way to get between North

53:11

and South Brooklyn without paying $65

53:13

and someone was just like

53:17

well how about these alternatives

53:20

that you can definitely get you between North

53:22

and South Brooklyn for less than $65

53:24

and it's like the B34 bus and the G train

53:30

so it's like you know that that's just kind of

53:33

like you know reinventing reinventing

53:35

the wheel and calling it a tech startup

53:37

is kind of like the thing also

53:40

I was looking through my

53:42

deleted email inbox just

53:44

to give some examples of like the type

53:47

of garbage that I'm getting one

53:49

of these things says AI girlfriends

53:51

are on the rise could they

53:53

make men behave better much

53:58

about absolutely not But they

54:00

will... It's like the general sense

54:02

of this is like, have you

54:05

done any thought about what this

54:07

problem is and that you're trying to

54:09

address? Like, why does it exist? And

54:12

that goes back to the age old, the whole

54:14

technology will save us thing. Like, we don't have

54:16

to address human problems because we can just make

54:18

it apps, which is one of my long term

54:20

just shakes in any space. Yeah.

54:23

It's just so apparent that even

54:25

a lot of the stuff that we're talking about with AI and

54:28

what have you, like at the core of

54:30

that is like economic anxiety, is lack of

54:33

education and social ills. And we don't want

54:35

to touch that. So

54:37

your inbox looks full of

54:39

AI girlfriends, Janice. Oh, yeah, for

54:41

sure. I think it's really responding

54:44

to the fact that, yeah, that people

54:47

are very anxious about the

54:50

economy for good reasons. And it's because

54:52

there's kind of like the money is

54:54

kind of gone, right? It's like people

54:56

are losing jobs, entire

54:59

industries are kind of like

55:01

dying out. It's

55:03

very tempting to basically say,

55:06

hey, we found a way to fix this

55:08

complicated social problem with

55:10

an app. And I kind of almost

55:12

understand why a lot of people are

55:14

able to accept that. It's sort of in the same

55:16

vein as the sort of mega

55:18

church preacher saying, I have

55:20

this thing, this one way.

55:23

It's kind of, you know, and that's like we've

55:25

talked about Silicon Valley as kind of like this

55:27

new kind of almost religious institution. And

55:29

it kind of resembles that in a lot

55:32

of ways. It's offering people very

55:34

simplistic solutions to complicated problems

55:36

that most people don't have

55:38

time to like fully think

55:40

about or analyze. People

55:44

are drawn to that because they still want to have hope. You

55:46

know, that's fair. That's understandable, even

55:49

if it's false. And

55:51

even if ultimately the reality is we've just got much,

55:53

much bigger work to do than AI girlfriends.

55:56

Yeah, your girlfriends are going to distract us from the

55:58

work we have to do. I think

56:00

that's a lovely place to close

56:02

out the conversation, unless Janice certainly

56:04

has something else. No,

56:08

I'm good. I'm going to go with a bunch of copyright

56:10

lawyers talk about, talk

56:12

about all this next. Thank

56:15

you so much for coming on to cyber and walking us through

56:17

this. It's

56:19

been such a pleasure. It's really been

56:21

a good conversation. Yeah, we're happy to

56:23

come back again. Thank

56:32

you. Thank

57:01

you. Are

57:54

you ready to enhance your future

57:56

in tech? Then it's time to

57:58

make your move to the Uk,

58:00

the nation that has more tech

58:02

unicorns than France, Germany, and Sweden

58:04

combined The nation that was third

58:06

in the world to have a

58:08

one trillion dollar tax sex of.

58:11

The nation were great. Talent

58:13

comes together. Visit Gov.uk forward/great

58:16

talent to see how you

58:18

could work, live, and move

58:20

to the Uk.