0:00

So I don't know how we haven't already used this prompt

0:02

mic. Perhaps we should be feeding

0:04

all of the podcasts into chat GPT to,

0:06

to tell us. But, um,

0:09

I'm going to ask you today, the prompt

0:11

that you get on chat GPT when you go onto the

0:13

site. And that is what can I help with?

0:17

Can you find a thousand or so

0:20

people who would like to purchase my social

0:22

media card game?

0:25

What? Yeah. 1 billion users. Let's get

0:27

the plugin.

0:28

Yeah, 1 billion users, which I will note

0:30

this week OpenAI announced

0:32

that they are aiming to get 1 billion

0:34

users on chat GPT.

0:37

So this all ties together

0:38

Oh my God. It's, do you think they, they took

0:40

inspiration from the Kickstarter?

0:42

They must have, but I would like them to

0:44

send 1 billion users to the Kickstarter.

0:46

We are not quite at our threshold

0:49

of support to actually make the game

0:51

go into production. But if you are

0:53

listening to this, I know you will like

0:55

this game because it is all about

0:57

social media and it is fun for the whole

1:00

family to play. So if

1:02

you're listening to this and you have not backed our

1:04

1 million users, 1 billion users.

1:07

Game on Kickstarter. You can just go to Kickstarter

1:09

and search 1 billion users. Please,

1:11

please check it out. We need more backers.

1:13

And if you know other people who are

1:15

interested in online speech and trust and safety,

1:18

and you want to have a fun game to play that

1:20

will also help you explain trust and safety

1:23

and social media and all this kind of stuff. Please,

1:25

please do it. I hate to beg, but I need to beg.

1:29

I'm, I'm going to buy five more.

1:32

There we go. We have a discount. If you buy five,

1:34

we have a discount,

1:35

I'm going to do it.

1:36

but anyways, Ben, what can I help

1:38

you with today?

1:39

What can you help with? I am incredibly

1:42

jet lagged, Mike. I've just landed back

1:44

from your fair nation after a week

1:46

of work there. And I do not know

1:49

what day or time it is. Um, I

1:51

have not figured out how,

1:53

To, yeah, to actually

1:55

speak as you can tell. Um, so any,

1:57

any advice, any help on how to get

2:00

back to a normal rhythm of sleeping would

2:02

be much appreciated. Let's see how

2:04

this goes.

2:05

Yes, yes, this, this may be an interesting one.

2:07

Uh, I can confirm that. Well,

2:09

one, I was up late last night working

2:11

on stuff and I was texting Ben who was

2:13

in Chicago. this was not

2:15

that many hours ago

2:17

No,

2:18

and now he is. In London,

2:20

and he showed me his, bag

2:23

that he had just thrown in the corner after

2:25

getting back home in

2:27

time to record this. So we are both

2:29

working on quite little sleep

2:32

and, uh, Ben even less than I. So

2:34

this may be a very fun

2:36

or very strange episode, depending

2:38

on where things go. Bye.

2:40

everyone. Hello

2:49

and welcome to control all speech, your

2:51

weekly roundup of the major stories about

2:54

online speech. Content moderation

2:56

and internet regulation. It's December

2:58

the 6th, 2024. And this week's

3:00

episode is brought to you with financial support from

3:02

the future of online trust and safety fund. And

3:05

by today's sponsor internet society,

3:07

a global nonprofit that advocates for an

3:09

open globally connected, secure,

3:12

and trustworthy internet for everyone. I'm

3:15

Ben Whitelaw. I'm the founder and editor of everything in moderation.

3:18

at least I, I am, or was

3:20

the last time I checked, um, and I'm

3:22

with an equally kind of bleary

3:24

eyed Mike Masnick, for altogether

3:26

different reasons. It's good to

3:28

be here, Mike. We haven't been in this, in these respective

3:31

chairs for a few weeks.

3:32

Yeah, that's right. We took off last week.

3:35

We missed all the big stories, uh,

3:37

during Thanksgiving week and then the week

3:39

before that you were traveling. So you were not

3:41

here. And so, uh, it's been, been a

3:43

little while since, since we've, been able

3:46

to do this together.

3:48

Um, we, we've got a lot to cover today, partly

3:50

because yeah, we were, we took a break last week.

3:53

happy Thanksgiving again for, to you and to our

3:55

listeners. And we have a really

3:57

good, bonus chat at the end of today's episode

3:59

with some super smart folks from the internet society,

4:02

around social media bands in Canada

4:04

and Australia and age verification in general.

4:07

So, stick around for that at the end. Thanks.

4:09

Today's show, and we won't cover those kinds

4:11

of stories today, just because we do that

4:13

really well with, with those folks. we're going to dive

4:15

straight in to, um, what is essentially

4:17

Mike a breaking story. Um,

4:19

it's, it's a live story that you

4:22

essentially woke up, rubbed

4:24

your eyes and, uh, have

4:26

dived into, it's the,

4:28

uh, the tick tock case it's back

4:30

on the agenda.

4:32

Yeah. So. We knew that

4:34

there was supposed to be a ruling coming down basically

4:36

any day. This is the, US law

4:39

that in theory forces

4:41

ByteDance to divest of TikTok

4:44

in the US or

4:46

forces, the app

4:48

stores and ISPs to block

4:50

access to it. And they

4:53

are supposed to either divest.

4:55

or we'll have access blocked on

4:57

January the 19th. And there's a whole sort

4:59

of political side

5:02

story to this, which is that our

5:04

incoming returning

5:06

president, uh, Donald Trump,

5:09

who originally was the one who proposed

5:11

the TikTok ban and tried to do so in a

5:13

very ham fisted way that failed, four

5:16

years ago. Has now

5:18

completely flipped positions, perhaps because

5:20

he was backed heavily by

5:22

one of, ByteDance's

5:25

biggest investors, but has

5:27

insisted that he's going to protect TikTok. But

5:29

he comes into office, I

5:31

think a day or two after this

5:34

law goes into effect. And so

5:36

TikTok had challenged the law and it went to the

5:39

DC circuit, which is in the appeals court.

5:41

And they came down with their ruling

5:43

today, basically saying

5:46

that the law is. Just

5:48

fine. There is no problems with the law

5:50

and TikTok must obey

5:52

it. And ByteDance must obey it. this

5:55

ruling did come out literally,

5:57

about an hour before we were recording.

6:00

I have read what I think

6:02

are the most salient points

6:04

in it. however, given the

6:06

timing, I may make some mistakes in

6:08

terms of this because I have not had a chance to go through

6:10

it in great detail.

6:11

Have you even had a coffee Mike? Are

6:13

I have not actually, I am,

6:16

I am, uh, yeah,

6:18

anyways, so, I

6:20

think they're wrong is,

6:23

is, is my quick summary. it's

6:25

interesting because the government tried

6:28

to argue that there were no first amendment

6:30

issues at all. And that the case could be decided

6:32

without even considering the first amendment.

6:35

The court went in a different direction and said that

6:37

government was wrong, but that the first amendment. Does

6:40

apply here. And then once you

6:42

say that the First Amendment does apply, then there's a question

6:44

of which level of scrutiny, and

6:46

there are different levels, you know have

6:49

to be passed in order for the law to be considered

6:51

constitutional. The highest bar

6:53

is is strict scrutiny. And

6:55

then there are, there are lower bars. And the government had argued

6:58

that if the First Amendment does apply, you

7:00

should go for one of the lower bars. ByteDance

7:03

had argued that it should be strict scrutiny,

7:05

which is the hardest to pass. The

7:07

court notes that there isn't

7:09

a comparable situation that says

7:11

which of the levels of scrutiny should

7:14

apply. And so they actually choose

7:16

not to, say which level

7:18

should apply. Though there is a concurring, uh, opinion

7:21

by one of the judges suggesting,

7:24

which level should apply, but

7:26

the court says it doesn't matter because

7:28

we think that this law passes even

7:30

the highest bar strict scrutiny.

7:33

And so then they just analyze it based on strict

7:35

scrutiny. And I think they do it.

7:37

in my one read through it, I

7:40

think they do a terrible job of it. They

7:42

basically buy into strongly

7:44

buy into the claims by the

7:47

U. S. government of the national

7:49

security concerns, even though

7:51

they admit that no actual evidence

7:53

is presented for those, but they basically

7:55

say, well, the U. S. government, a lot of people

7:57

in the U. S. government say that we should be concerned

8:01

and therefore we're just going to take that as fact.

8:03

This is not uncommon in

8:05

the U. S. court system when

8:08

the government says, well, we have national security and

8:11

blah, blah, blah. We can't tell you what they are. We're just

8:13

really concerned. the courts are often

8:15

willing to go along with that. That

8:18

has led to a whole bunch of really terrible things

8:20

having to do with civil liberties and civil rights

8:22

and all sorts of stuff going back

8:24

decades. But that seems to happen

8:26

again here where they're just like national security concerns

8:29

seem totally legit and you

8:31

know ByteDance hasn't given us anything to

8:33

Reject that they admit multiple times

8:35

that a lot of these concerns are totally speculative

8:38

And yet because ByteDance can't respond

8:40

to the questions Speculative concerns.

8:43

Therefore, this is all okay. There's

8:46

a whole bunch of other like little stuff in here that

8:48

again, it's just kind of like, well,

8:51

you know, by chance I said, we have all these other,

8:53

less restrictive means of doing this, which is part

8:55

of the strict scrutiny test. And the court

8:57

is like, nah, you know, we don't think any of those

9:00

good enough. I mean, you know, Bytance really pushed

9:02

this idea that, they would separate out

9:04

all the operations, which they've mostly done.

9:07

Everything is hosted in the U S by Oracle.

9:09

They've given Oracle the power to audit stuff.

9:11

they even offered the U S government, like an off

9:14

switch, like that they could, you

9:16

know, push a button and turn off TikTok, you

9:18

know, in the case that something really bad happened

9:20

and the government was like, nah, you know,

9:23

that doesn't really satisfy what the government

9:25

is doing here. There are a bunch of other oddities

9:27

in here. You know, there were issues around

9:29

like specifically targeting, like the bill actually

9:31

names Tik TOK, which seems

9:34

like it's singling them out for different treatment. And

9:36

the court kind of waves that off

9:38

and says, well, no, that's

9:40

not really true, even though it does mention

9:42

it. it, they said at one point, it's not

9:45

punishment. If it was a bill of attainder, which is

9:47

a specific thing where it's like just targeting someone

9:49

for punishment, they said, yes,

9:51

it only names TikTok, but not for

9:53

punishment. And therefore it's not

9:55

a bill of attainder. There are all of these like

9:58

really odd things. So

10:00

the big question now is kind of like what happens

10:02

next? Like does TikTok.

10:04

get shut off on January 19th

10:07

and be gone for two days. And then Trump

10:09

comes into office and suddenly

10:11

changes position and allows to talk to come

10:13

back. I don't know. I mean,

10:16

I, I think, and

10:18

I could be, there may be, there's some procedural

10:20

weirdness in terms of how this law was written

10:23

in terms of like forcing it to go straight to the

10:25

circuit court rather than a district court, which would

10:27

be the normal thing. I assume

10:29

that they can now go to the Supreme Court and

10:31

ask for a stay on this ruling,

10:34

in order to appeal it. And maybe that puts

10:37

the decision off past, January

10:39

19th and allows Trump to get into office

10:41

and then sort of, you know, do something

10:43

to keep TikTok, unless he decides that he,

10:46

he maybe doesn't want TikTok. We, we

10:48

don't really know at this point.

10:50

Who knows what side of the bed Donald Trump will,

10:52

will wake up on. And so it's

10:55

really interesting to me, Mike, thanks for unpacking that. It's really

10:57

interesting to me that all of the work done by TikTok

11:00

to essentially kind of separate

11:02

it out as a, as a. distinct

11:04

legal entity, under this kind of project,

11:06

Texas, umbrella that it, that

11:09

it came up with a few years back actually has seemingly

11:11

not worked at all. it has not proven to

11:13

be persuasive to, in this case

11:15

at all as to the national security,

11:18

threat still. Why do you think that

11:20

is, why do you think that the, the kind of, All

11:22

of that efforts has been, has been kind of fallen

11:24

on deaf ears,

11:25

I mean, there is some element of it, which is

11:28

just, honestly, it feels like general

11:31

fear of China

11:33

and that comes through in the ruling to

11:35

like you see repeatedly, just

11:37

talk of the PRC, the PRC, this,

11:40

the PRC, that, and even

11:42

if they say they're not going to, if the PRC

11:44

comes down and demands that they do this or that,

11:46

they will have to obey and

11:49

therefore, that is the overriding

11:51

concern. And, you know, there were stories

11:54

that came out in the press from

11:56

like former TikTok employees saying

11:58

that, employees in China still

12:00

had access to data and

12:03

that that's noted in the decision. They,

12:05

they point that out. And so they say, you know, we just, basically

12:08

they just don't trust that this is real.

12:10

They, it's sort of. the judges seem to

12:12

feel that this was kind of a fictional setup,

12:14

um, and they don't really trust

12:16

that Oracle's ability

12:19

to audit it as actually meaningful

12:21

because again, they, they feel that, the PRC

12:23

could come down and, do all of this. You

12:26

know, the other argument that is made too, is that,

12:29

It's really just focused on the PRC.

12:31

I mean, because they say,

12:33

you know, one of the reasons why they claim that this isn't

12:35

a first amendment violation is because if

12:38

the company does divest, they

12:40

say all of the same content

12:42

and all of the same moderation could still

12:44

occur. There would be no change to

12:46

the, expressive nature of things. The only

12:49

thing we are doing is trying to

12:51

disconnect the PRC from that. From tick

12:53

tock. And so they feel that,

12:55

the project Texas set up and

12:57

the Oracle audits and the U S government

12:59

like off button don't separate

13:02

the company from the PRC. And so

13:04

that's where the court comes down.

13:05

They'd only be happy if, the PRC

13:08

sold by dance or by dance,

13:11

changed owners or changed hands. That's, that's

13:13

the only way this would actually kind of, uh, change

13:15

the, change the outcome in, in many sense.

13:17

Yeah. And, and, you know, I think for a variety

13:19

of reasons, I think, I think it's wrong. I think

13:21

it's, there are other precedents

13:23

on this that they sort of poo poo. And

13:26

we're just like, the hat doesn't, doesn't really apply here.

13:28

but it is something that happens. I think, I think

13:30

it's a, bad look. I think it's a bad look for, you

13:33

know, a supposedly free country in the U

13:35

S to sort of take this viewpoint. And

13:37

I think it will justify all sorts of other bad stuff

13:40

from elsewhere as well, based on

13:42

things like this, uh, it's, I think it's

13:44

a bad look. I know some people really.

13:46

Are concerned and there may be legitimate reasons

13:49

to be concerned about the Chinese government

13:51

and their connections here But I just

13:53

feel that this law goes against, Basic

13:56

american principles on on things and

13:58

and i'm disappointed by by the ruling which you

14:00

know again I've only read it once and really focused

14:03

on the first amendment section, which is sort of the second

14:05

half of the ruling And, um,

14:07

I just, uh, it strikes me as very unconvincing.

14:10

yeah. And that national security concern

14:12

has yet to be proven as real

14:15

or anything beyond really hypothetical at this stage.

14:17

Yep. Yep. And, you know, it, it, that frustrates

14:20

me. I mean, I have this come up in all sorts of, other

14:22

cases around like Fourth Amendment stuff and

14:24

encryption and just, you know, the willingness

14:26

of the courts to accept the government just sort of

14:29

giving this blanket statement that this is a national

14:31

security concern without proof is,

14:33

is just kind of frustrating.

14:35

Yeah. Okay. it links, um,

14:37

nicely to, to our next story, Mike, which

14:39

you have, Figured out, um,

14:43

it's, it's a TikTok story, but it's about a country

14:46

that we don't often talk about in controlled speech, um, Romania.

14:49

and I guess this is kind of really what the, in

14:51

some senses, the threat that

14:53

the, courts, and the government in

14:55

the U S are trying to kind of mitigate against. Right.

14:57

Yeah. Yeah. It's, it's kind of

14:59

an interesting, interestingly related.

15:02

Um, there's an election

15:04

going on this coming Sunday, in

15:06

Romania, but in the first

15:08

round of the election, that's sort of a, you know, a

15:10

two round election, the first round to figure

15:12

out who are the top two candidates and then the,

15:14

final election is

15:16

you know, with just those two top two candidates,

15:19

there was a surprise, you know, second place

15:21

finish that knocked out the incumbent.

15:24

The incumbent, I believe, came in third and

15:26

so is not taking part in the election

15:28

this weekend. there was a candidate,

15:30

and I may pronounce his name wrong, but Kalin

15:33

Georgescu, who was

15:35

considered a sort of mostly unknown

15:38

far right. populist

15:40

candidate who sort of came out of

15:42

nowhere with a big TikTok following.

15:44

Um, and he has been

15:47

sort of very pro Russia,

15:49

pro Putin. And in fact, a lot

15:51

of his successful TikToks are

15:53

sort of reminiscent of,

15:56

Putin. Style, uh,

15:59

press appearances, riding

16:01

a horse, doing judo,

16:03

apparently running on a track without breaking

16:05

a sweat. It's just sort of this like physical

16:08

prowess.

16:09

nothing, nothing to, you know, who doesn't like

16:11

that? That sounds like a great, a great TikTok

16:13

experience.

16:14

yes. Yeah. And so he built

16:16

up a really big following on, Tik

16:18

TOK. The polls in the country had suggested that

16:20

he was, not going to get that many

16:22

votes. and a lot of people, it

16:24

appears sort of chalked up his. Big

16:27

Tik TOK following to

16:29

foreign influence campaigns. They also noted that those

16:31

videos were like highly stylized

16:33

and produced. And there were questions

16:36

of who was doing some of the

16:38

producing and the stylizing behind

16:40

it. There were questions about, whether

16:43

or not the followers were real, were

16:45

they foreign efforts? Were

16:47

they bots? there were questions

16:49

around. Influencers

16:51

who were promoting the videos

16:53

and were they paid or not?

16:55

I, I, I found this really interesting

16:57

point around, like, whether this candidate was, marked

17:00

in the same way as other political candidates, basically

17:02

suggesting that, they weren't kind of tagged

17:05

as a political candidate running an election

17:07

and therefore gotten larger reach

17:09

than the other candidates, which is a really interesting

17:12

kind of element to this, isn't there?

17:13

Yeah. And that's unclear because tick

17:16

tock denies that. but deal

17:18

is because there are restrictions on

17:20

political advertising within

17:22

Romania, If all the candidates

17:24

are treated as candidates, then

17:26

in theory, there were limitations on kind of what sorts

17:28

of promotions they could do. And if they were paying

17:31

influencers to promote, is that

17:33

a political ad? and how does that apply?

17:36

And so there are all these questions and the suggestion

17:38

from many, including some

17:40

government officials is that TikTok

17:42

didn't designate him as a, politician,

17:44

as a candidate, and therefore didn't have

17:47

these restrictions and that allowed his videos

17:49

to go much further and build up a bigger

17:51

following. Again, TikTok denies

17:53

it. So it's a little unclear whether or

17:55

not that is true. There is though, that,

17:58

the things that TikTok. Did do

18:00

it was at one point they did remove

18:02

what they refer to as a covert network

18:05

that was trying to boost

18:07

his videos. And so, they

18:09

did. Something, you

18:12

know, they certainly, paid attention to something

18:14

and found some inauthentic

18:16

behavior behind him and that they

18:19

claim that they stopped. And they also said that

18:21

they found some similar types

18:23

of inauthentic behavior for some of the other candidates.

18:26

Um, so it wasn't that they did nothing.

18:29

they were willing to step up and do

18:31

something, but Romanian

18:33

officials, feel pretty sketched out by

18:35

this and they've asked the EU to investigate.

18:38

and you know, it, it brings up all the questions

18:40

that happen when, whenever there are

18:43

unexpected results in elections

18:45

and everyone's looking for explanations and, and

18:47

these days, oftentimes they quickly

18:49

jump to, well, it was this internet platform

18:52

that caused the problem. Yeah.

18:53

Yeah. I mean, there's a couple of interesting points

18:55

here is in some ways this story is, one

18:57

we've seen many times before, which is

18:59

platform being used by

19:01

political candidate. with the kind of

19:03

looming threat of foreign interference in the background.

19:06

And, you know, we've there's countless examples of that,

19:09

Analytica being the kind of most, most famous

19:11

one, I guess. And, It's

19:13

nice to see that things don't change in the

19:15

best part of a decade. But

19:17

I was really interested in, you know, if that is the case, if there

19:19

is a kind of lack of oversight here, Romania

19:22

is pretty small. It has 20 million citizens.

19:25

8 million of those apparently use

19:27

TikTok or have a kind of TikTok account. So it's a significant

19:30

number. You know, why

19:32

was it that there wasn't really, I guess, more attention

19:34

paid to Making sure that candidates

19:37

were, given the same kind of platform. And

19:39

I decided to kind of go into the DSA

19:42

data, my find out how many,

19:44

how many moderators Romania has,

19:47

um, or how many are Romanian speaking. And,

19:49

and so there are, according to the DSA

19:51

report from January to June,

19:54

95 Romanian

19:56

speaking moderators for a country

19:58

of 20 million. And we don't

20:00

know if that's good or bad. It doesn't necessarily take

20:02

into account, I guess, people setting policy

20:04

either. That's probably just the folks who are, looking

20:06

at reports and appeals and those kinds of things.

20:09

But it doesn't seem a lot compared to other countries.

20:11

so the Netherlands, which

20:13

has roughly the same number of. citizens,

20:16

we don't know exactly how many users, of

20:18

TikTok, but they have 160 moderators,

20:21

speaking Dutch.

20:22

roughly double,

20:23

So roughly double, um, Sweden

20:26

has, roughly the same number of moderators,

20:29

99, but has half the population. And again,

20:31

we don't, we don't know how many, Of those people

20:33

use, use Tik TOK. So again,

20:36

you know, it's interesting to me

20:38

that kind of how resources are

20:40

applied. And again, this is a tale

20:42

as old as time, how platforms,

20:45

in the kind of non core markets,

20:48

set policy and operationalized

20:50

policy is something that we, we see time and time

20:52

again as, as a sticking point, do

20:54

you think that's kind of relevant here? Does that feel like a,

20:56

a pathway trod a little bit?

20:59

Yeah. I mean, you know, it is one of these,

21:01

big questions that comes up all the time, so

21:03

many of these discussions, you know, and we

21:06

try to, you know, one of the reasons

21:08

you and I are always looking for stories outside

21:11

of, the U S and, The

21:13

big countries that everybody talks about is because

21:15

there is an important story there about how

21:18

these companies handle it and the, you

21:20

know, less followed countries where,

21:22

where less attention is, paid to them. And

21:25

so I think, I think it is a, it's a huge

21:27

story. And, you know, there is this element

21:29

of this one where I sort of feel like, I

21:32

think. A lot of people really believe

21:34

that so much of his following, in

21:36

this case, were bots, that

21:38

they didn't think that the votes would follow, and

21:40

yet the votes did. And so, there was

21:42

this part of me when I first story where I was

21:44

like, and, you know, the first thing I was reading

21:46

was like, oh, you know, it's all fake followers

21:49

and bots. I was like, yeah, but it wasn't fake

21:51

voters, you know. So,

21:53

so something is happening here. but

21:56

to be honest too, like, I don't know that, you can

21:58

say necessarily that like, Oh,

22:01

it's 95 or whatever moderators

22:03

too little. I don't know. And

22:05

we, we don't know, like, was

22:07

that not enough where the policy's not in place again,

22:10

there's the whole thing where like TikTok has denied this

22:12

and they did take down some inauthentic

22:15

behavior. so it, we, we have,

22:17

you know, Partial information, not full information.

22:19

If the EU does an investigation,

22:22

it would be really interesting to find out if

22:24

more details come out of it, but it is,

22:26

it is sort of an interesting story pay attention

22:28

to and to see what comes of this.

22:30

yeah, definitely. And I mean, I think the other part of

22:32

this story, something that kind of is

22:34

a thread that runs through U. S. TikTok ban

22:37

as well, is just the, is

22:39

a media story, really. It's about the fact that

22:41

there is so many people kind of

22:43

consuming TikTok and the shift

22:45

away from traditional media to, new

22:48

forms of media. The fact that politicians can bypass

22:51

traditional forms of media. Yeah. as this

22:53

candidate has done and still perform

22:55

very well in an early round. So we'll

22:58

keep a tabs. I've never said this before, Mike, but I'll be keeping

23:00

tabs on the Romanian election this Sunday. Uh,

23:03

There you go. This, this, this coming Sunday

23:06

is my birthday. So I will celebrate it by paying

23:08

attention to the Romanian election as well.

23:11

happy birthday in advance. And, uh,

23:13

you.

23:13

What, what a way to celebrate

23:15

Yes, yes.

23:16

so if those kind of two stories paired

23:19

together, I guess, represent in some ways the

23:21

kind of ghost of, antitrust past

23:23

Mike, you know, the way that the,

23:26

a segue. What a segue.

23:28

you know, I think the next story is the ghost

23:30

of, of antitrust future. next one

23:32

you picked out and it's about a man that I actually

23:34

didn't know about, uh, up until,

23:37

an hour or so ago. Um, but

23:39

he's an incredibly frightening looking man. I'll have

23:41

to say, and I'm not sure if I'm ever going to forget

23:43

his face. Um, so

23:45

tell us about Andrew Ferguson and,

23:47

and what he's come out with this week.

23:49

Oh my goodness. So Andrew Ferguson is an FTC

23:51

commissioner. and as

23:54

a Republican, the FTC

23:56

has five commissioners, three are

23:58

appointed by whichever party

24:01

has the presidency and two are the,

24:03

other party. So right now there are two

24:05

Republican commissioners. FTC commissioners and

24:08

three Democratic ones, that will flip

24:10

in January. And,

24:12

Andrew Ferguson is one of the Republican

24:14

commissioners. He is vying

24:17

for very clearly vying for

24:19

the chair, taking over what Lena

24:21

Khan's position is, is right now under

24:23

a Trump administration. and

24:26

there was a New York Post article Today

24:28

or yesterday that sort of detailed, there are three

24:30

leading candidates, the two current FTC

24:32

commissioners, which is Ferguson

24:35

and the other one's Melissa Holyoke.

24:37

and then there's, there's a third person,

24:39

who, worked for Senator Mike Lee

24:42

that he's really pushing for to be chair. And

24:44

the, the question that people around

24:47

Trump are apparently asking is like, which

24:49

one of these is going to be toughest on,

24:52

they say, big tech. but they really mean

24:54

Trump's enemies. Especially considering

24:56

how much support he got from certain tech

24:58

sectors this time around, it'll

25:00

be, you know, the companies that they don't like.

25:03

And so the FTC took a

25:05

fairly typical FTC action

25:07

this week against an e commerce platform

25:09

called Goat. The details aren't even

25:12

that interesting in the case, but basically

25:14

they lied about shipping times.

25:16

People were paying for premium shipping and not getting

25:19

it in time. And then also

25:21

they had a, buyer protection, thing

25:23

where they're saying like, if anything goes wrong, we'll protect you. And

25:25

then they weren't, they weren't living up to that. The

25:28

FTC took action on them and basically saying,

25:30

you know, they were making promises and they weren't living

25:32

up to it. That's an unfair and deceptive practice.

25:35

Very typical standard FTC.

25:37

Nothing at all interesting about that. The

25:40

other Republican commissioner, Holyoke,

25:43

put out a concurring statement on this

25:45

that basically just said, this also

25:48

proves that, we

25:50

can use the same authorities to go

25:52

after big tech companies for

25:55

unfair moderation decisions.

25:57

Okay.

25:58

Which is nonsense.

26:00

Yeah.

26:00

but it was like a one paragraph thing

26:03

and it looks like to me, at least

26:05

Andrew Ferguson then said, okay, I see

26:07

that I got a one up it because we're

26:10

in, we're in a fight here for the chair position.

26:12

I raise you.

26:13

I raised you, I raised you crazy.

26:16

and put out this like four page concurring

26:18

thing. Again, none of this has anything to

26:20

do with the ruling on goat, which is the company

26:23

that this is ostensibly about, saying

26:25

like, yes. And I agree with Holyoke that

26:27

we can. Use this power to

26:29

go after unfair moderation, but

26:31

also we, we

26:34

can use our powers to go after

26:36

advertisers who stopped advertising

26:38

on X because that must

26:40

be antitrust collusion

26:42

to censor free speech. And

26:45

we have to support free speech. And

26:47

right now there's only one free

26:49

speech platform Elon Musk,

26:51

the brilliant, wonderful free speech supporter.

26:54

And, you know, How dare,

26:56

any other platform censor

26:59

American speech and that must be illegal

27:01

and how dare they not advertise. so

27:03

he goes after, advertisers who

27:05

stopped, he goes after Garm, which we've talked

27:08

about in the past saying, you know,

27:10

that was, clear evidence of, of

27:12

collusion. He goes after NewsGuard,

27:15

saying that, NewsGuard, who I've written about

27:17

a few times that the Republicans have gone crazy about,

27:20

NewsGuard, all they do is just say which

27:22

news sources are trustworthy and which

27:24

are not. And he sort of

27:26

admits when he's talking about NewsGuard, like,

27:29

yes, NewsGuard can have its own opinions. But

27:31

if multiple companies are basing

27:33

decisions on those opinions,

27:35

that is antitrust collusion. It,

27:38

it is four pages of crazy

27:42

indicating once again, like

27:44

with Brendan Carr, we talked about a

27:46

few weeks ago, like with Brendan Carr

27:49

that, these. Bureaucrats

27:52

really intend to use the powers of government

27:54

to attack speech online,

27:57

and they're framing it all within

27:59

the language of free speech. The whole

28:01

thing over and over again, he talks about,

28:03

how important free speech is. And he does

28:06

the Elon Musk is the only believer in

28:08

free speech. And every platform

28:10

has to use the same policies

28:12

as, Elon does. And how

28:14

dare they not do that?

28:15

so just to clarify, so we have a situation

28:17

where we're antitrust as we've

28:20

talked about in the last couple of weeks of the podcast

28:22

which is going to be a big theme within the Trump administration,

28:25

and, you know, he's run out of the FTC

28:28

has a guy in Brendan Carr

28:30

who doesn't know much.

28:33

Brendan Carr's the FCC, not the

28:35

Sorry, sorry. Yeah, so, so the

28:37

FTC commissioners also

28:39

don't know very much according to this letter,

28:41

at least this particular guy. And

28:43

so we have a situation where like maybe

28:46

nobody knows anything about

28:49

it's, I mean, the question is, do

28:51

they know or are they just like putting on

28:54

a show for Trump? Right. And, and

28:56

it's just this sort of like populist thing.

28:58

I don't know Andrew Ferguson that well, Brendan

29:01

Carr, I know a little bit. And so I know

29:03

he knows that he's. Lying,

29:05

like Brendan Carr is smart enough to

29:07

know what the law is. I don't know enough

29:09

about Ferguson to know whether or not he knows

29:11

this is crazy. You know, one

29:13

of the lines that really got me in this letter

29:15

was like, he claims at one point that the proof

29:18

of collusion among big

29:20

tech companies to censor content

29:22

in an illegal manner is

29:24

that simultaneously,

29:26

he specifically says simultaneously,

29:29

all of the big tech platforms

29:31

blocked. All discussion

29:34

and reporting on the Hunter Biden

29:37

laptop in 2020.

29:39

Which just isn't the case.

29:41

None of that happened, right? The

29:43

only thing that happened was

29:45

two companies took some action, Twitter

29:47

and Facebook. The action that Twitter

29:50

took was it blocked the link. It did

29:52

not block any other reporting on it. There was

29:54

other reporting on it. It did not block any

29:56

discussion of it. There was lots of other discussion

29:58

on it. In fact, like it was like a trending

30:01

topic. The only thing they did was

30:03

they blocked the link. To a single

30:05

New York post story for 24

30:07

hours. Then they reversed their policy

30:09

and allowed that link to be shared.

30:12

The only thing that Facebook did was it said,

30:14

well, there's some questions about the story, so we're going to

30:16

keep it out of the trending. It won't

30:18

go into the trending topics. and

30:21

they reversed that policy relatively quickly.

30:23

That was the only thing it did, but he declares

30:26

unequivocally that the entire. Tech

30:28

industry simultaneously blocked

30:30

all discussion of this. And

30:33

I, you know, one of the things that gets me

30:35

is I pointed out earlier this year that Elon

30:38

Musk did all that and more when

30:41

apparently Iranian hackers got

30:43

access to the Trump campaigns dossier

30:46

on JD Vance and they

30:48

passed around and most, most media sources

30:50

didn't bite on it. Finally, Ken Klippenstein,

30:53

who has a substack posted

30:56

it and Elon

30:58

banned Ken. He blocked

31:00

all. Links to,

31:02

any part of Ken's sub stack, not

31:05

just that one article. He,

31:07

you know, pulled down all sorts of stuff. And to this day,

31:09

I don't think you can share that. He did let

31:11

Ken back on the platform after like

31:13

two weeks, So everything

31:15

that, they have accused Twitter of

31:17

doing to the Hunter Biden laptop story, Elon

31:20

has done and more and gone much

31:22

further. And yet in this. comment

31:25

from this FTC commissioner, he

31:27

claims that Elon is the big free speech supporter

31:29

and the actions taken on the, the

31:31

Hunter laptop, which didn't happen,

31:34

prove that they're illegal

31:36

censoring, collusion, antitrust.

31:39

Not everything about this.

31:41

You've got your hands on your head, Mike. It's

31:45

it's so wrong, but this is,

31:47

this is unfortunately the world that we're living in. and

31:49

it, gives a sense of how the

31:52

incoming administration is going

31:54

to attack content moderation.

31:56

They're going to make these claims. They are

31:58

going to try and use every legal lever

32:00

they have, even as they are crazy

32:02

and totally counterfactual to reality.

32:05

Yeah. There's no I mean, the

32:07

idea that FTC could prove that

32:09

platforms coordinated on policy

32:11

changes or anything like that would

32:13

be so difficult to do, right? If this actually was,

32:16

you know, is a route you want to go down,

32:18

how do you go about saying that

32:20

this company over here would Has done

32:22

the same thing as this company over here a

32:24

way that amounts to collusion.

32:26

Yeah, well, the thing that they can do, and they

32:28

probably will do, is that they can conduct

32:30

investigations, and they can demand

32:33

to see all sorts of internal files, and

32:35

that is what is going to happen, almost certainly,

32:37

and then, I would guess what would come out of it

32:40

is, probably a, really

32:42

misleading investigation. findings

32:44

and they'll release things selectively that

32:46

take things out of context and make, make

32:49

claims that are just not accurate. and

32:51

it's going to be a mess. and this is why we're

32:53

seeing, tech companies, you know,

32:55

trying to kiss the ring of Donald Trump

32:57

and, and try and make nice because they know

32:59

if they don't, they're going to face all

33:02

of this kind of, authoritarian

33:04

nonsense.

33:04

yeah, no, indeed. And, and,

33:06

you know, currying favor, it's actually,

33:09

uh, again, a very good segue onto our next story,

33:11

um, is, is fast becoming

33:13

did it on purpose, Ben.

33:15

The theme of this, episode. So, you know, we

33:17

have a situation where, FTC

33:19

commissioners are, are, cozying up

33:21

to, the new administration. We also

33:24

have a situation where Meta

33:26

in a very coordinated way is

33:28

doing the same thing. and so this week, a number of

33:30

different outlets, including the financial

33:32

times and the verge reported on,

33:35

Nick Clegg, the, president of global

33:37

affairs at Metta, talking about how,

33:40

Metta essentially overstepped the line

33:42

when it came to content moderation during COVID,

33:45

the comments were made in a, reporter.

33:48

Briefing, which is, uh, you know, does

33:50

happen, but it's a very kind of coordinated,

33:52

very kind of controlled environment, for

33:55

a very senior person within Metta to

33:57

kind of make these statements. And, yeah,

33:59

the, coverage has been essentially

34:02

that Metta is not

34:04

apologizing, but admitting that it, it overstepped

34:06

the line in terms of, COVID information

34:09

controlled during that period. And. This

34:11

comes on the back of, you'll remember Mike, that letter

34:13

by Mark Zuckerberg, uh,

34:16

to Jim Jordan, which we had a good laugh

34:18

about, um,

34:21

I thought it was a cry.

34:23

we cried a bit, um, it, it,

34:25

it felt like a letter that had been,

34:28

written at gunpoint, um, I remember, I

34:30

remember saying, and, uh, almost

34:32

kind of made, made to write that, and it was clearly

34:34

in lieu of this situation, right, that where

34:36

Donald Trump becomes president again,

34:39

and, you have a situation as was reported

34:41

this week, where Zuckerberg is invited

34:43

to Mar a Lago to talk about the future

34:46

of tech policy, out of the U S. So in

34:48

a number of different ways we have, you know,

34:50

in here, we have meta kind of cozying up to

34:52

the new administration. We have the FTC commissioners

34:54

doing the same. Is there anybody that

34:56

has any dignity left? You

34:59

know, I think this stuff is so obvious,

35:02

right? It's, you know, it's so obvious and in some respects,

35:04

I'm frustrated about the way that it's

35:06

reported, in this way, because

35:08

apart from the line that says This is

35:11

a briefing with Nick Clegg and

35:13

some journalists who've been invited there. quite

35:16

clearly, you know, designed to be a signal

35:19

to the Trump administration of,

35:21

we know what you're going to ask us to do and

35:23

we're happy to do it.

35:24

Yeah. Yeah. That's exactly what it was. this

35:26

was totally a messaging thing. It

35:28

was, you know, a coordinated attempt by

35:31

the company to lay out this message that

35:33

will be embraced by. the

35:35

sort of MAGA faithful, to insist that it

35:37

proves, I mean, this will

35:39

be extended, right? They'll say it proves that,

35:41

not only that MEDA was

35:44

overly, willing to suppress speech,

35:46

but it will just be reinforced with the claim

35:48

that, the demands for that came from

35:51

the government, which is the part that

35:53

he didn't say that. But that is,

35:55

claim that lots of people are making. and

35:57

so, This is, it's a spineless

36:00

capitulation to this argument

36:03

and basically it's giving

36:05

the Republicans ammo to claim that

36:07

we were right all along. We were unfairly

36:09

targeted. We were unfairly censored

36:12

and, even meta admits

36:14

it. and we'll see that over and over again.

36:16

And people will point to this as if it's proof. I

36:18

had somebody yell at me this week,

36:21

only one.

36:22

well, there were a few people, but someone was yelling at

36:24

me about this, where I was talking about

36:26

some of this and they were saying, well,

36:28

you know. Zuckerberg admitted under

36:31

oath that the US government pressured

36:33

him to take down content. He didn't want to take which is

36:35

not What actually happened,

36:37

Right.

36:38

but like the message gets out there

36:40

and kind of knows what they're doing when they,

36:43

when they say this. and you

36:45

know, I understand why they're doing it. You know, they feel

36:47

like they need to do it to avoid,

36:49

to hopefully avoid costly stuff, but

36:51

it is, shows a real lack of principles

36:54

as far as I'm concerned.

36:55

Yeah. And if you're a trust and safety professional

36:58

working in meta who,

37:00

probably spent countless hours

37:02

trying to figure out what the policy

37:04

should be during an evolving situation

37:06

that no one has ever seen before that was COVID

37:09

and no one will ever see, you know, for a

37:11

long time, that's going to be really, really

37:13

tough to take.

37:14

it's demoralizing, right? I mean, you know,

37:16

the reality is what these companies should be doing

37:19

and they don't do is saying, like, these are really,

37:21

really, really difficult decisions.

37:24

And there was no way to get it right. There was

37:26

simply no way. I mean, I talk about sort

37:28

of impossibility theorem here. There

37:30

is no way to get it right. And that is it. Extra

37:32

true in a case where you have something that is brand

37:34

new. Nobody understands, you know, nobody understood

37:37

the details of COVID. We didn't know what

37:39

was right. We didn't know what was wrong. And

37:41

people made choices and lots of people made

37:43

wrong choices. Some of them made wrong

37:45

choices because they just didn't have enough information

37:47

and there's more information came out, they adjusted. Some

37:50

people made wrong choices because they, you

37:52

had crazy ideas in their head. There were

37:54

all sorts of wrong choices that were made along

37:56

the way. And it wasn't because of like

37:58

any, you know, in many

38:01

cases. It wasn't because of

38:03

bad actions or bad ideas. I think,

38:05

you know, the companies. Try to put

38:07

forth their best effort. That's what Meta should

38:09

be saying. You know, yes, we

38:11

may have made mistakes, but we made best efforts

38:14

based on what kind of information we had.

38:16

We took this seriously. We wanted to keep people

38:18

safe. And because of the changing

38:20

nature of the information environment, we

38:22

had to make decisions on the fly. And instead

38:25

he comes out and he gives this statement, which is basically

38:27

like, Oh, you know, we took down too much content

38:29

because there was too much pressure on us. Like, you

38:31

know, come on, stand up, have a spine. I,

38:33

it's. It's really, really

38:35

frustrating to me. This was a chance. This

38:37

was an opportunity for them to educate people

38:40

on how trust and safety works and what the

38:42

real purpose of trust and safety is. And

38:44

instead he's feeding into the narrative

38:46

that it's this awful censorship machine.

38:48

And it's, it's really, really frustrating.

38:51

Yeah. I mean, I've talked to a few

38:53

folks recently about trust and safety is kind

38:55

of marketing problem. The fact that, you know, it

38:58

needs to kind of present itself continually as

39:00

that difficult challenge, those impossible

39:02

trade offs, and it needs to kind of completely an

39:04

ongoing message that out, um,

39:07

it doesn't help when Nick comes out and gives a,

39:09

a soft briefing to those journalists that the

39:12

opposite is true. You know, he does say,

39:14

interestingly, just before we move on

39:16

about how, AI has

39:19

all this potential, but right now there are some really

39:22

pissed off people on, on Metas platforms

39:24

for the fact that, they make mistakes and

39:26

remove innocuous or innocent content. And,

39:29

again, you know, just. it's kind of completely

39:31

crazy because it was literally two weeks ago.

39:33

We were talking about how, how threads was an

39:35

absolute mess of a, of a moderation

39:38

process and you had all of these kind of, terms

39:40

that should have been moderated being taken down. So I'd

39:43

love for media to be a bit more critical

39:45

and, and, uh, clear on, on

39:47

what it is that senior people like Clegg are

39:49

saying, and I think that's, that's really part

39:51

of our job, um, to, to, to do

39:53

that too.

39:54

they, they, you know, put this in context,

39:56

like put, put his statements in context

39:58

and I didn't feel like, like the media

40:00

was really doing that.

40:02

no, no, indeed. Mike, I'm gonna,

40:04

um, we can't finish on that low note.

40:06

we've got a couple of lighter stories.

40:08

Um, but I'm gonna, throw to you to

40:10

pick, you had, end

40:13

with ChatGPT since that's where we started

40:16

episode. Um, tell

40:18

us about this kind of fun lighthearted story.

40:20

Yeah. This was kind of an interesting story where

40:22

suddenly it started spreading wide that, chat

40:25

GPT would break if you tried to get

40:27

it to say anything The

40:29

name David Mayer. I saw

40:31

it first in a very funny post

40:33

on blue sky from Andy Bayo,

40:35

who's is really interesting guy runs,

40:38

uh, has the site waxy. org. And

40:40

he does lots of really interesting stuff, but

40:42

he had heard that. And so he, he

40:44

created this question for chat GPT, which was

40:46

combined the first name of the artist who recorded

40:49

Ziggy Stardust and the last

40:51

name of the artist who recorded your

40:53

body is a wonderland into a single

40:55

name and chat GPT starts.

40:58

And says the artist who recorded Ziggy Sardust

41:00

is David Bowie. And the artist who recorded

41:03

Your Body is a Wonderland is John Mayer.

41:05

Combining their names and it says David,

41:07

and then it breaks and it says, I am unable

41:10

to produce a response. It

41:12

refuses to produce the name David Mayer.

41:15

And what people, yeah,

41:17

what people then quickly discovered was that there

41:20

was a short list of names that

41:22

chat GPT will break if you try

41:24

and get it to produce those names. Then

41:27

the hunt was on to sort of figure out who they were

41:29

and why

41:31

odd. So, and did they find

41:33

out what the cause was that, what, what, why doesn't

41:35

ChatGPT like these names?

41:37

so as far as anyone can

41:39

tell, and as I don't think open AI

41:41

has come out and said anything yet, and I will

41:43

note that they have fixed the David

41:45

Mayer one, so that it

41:47

now does work. But the other names

41:49

on the list do not work, still do not work.

41:52

Okay.

41:53

It appears that OpenAI

41:55

just created a block list. It

41:58

appears to be about six names. There might be

42:00

more, um, but there's a

42:02

block list that if you try and get it to produce that

42:04

name, it will break. And

42:07

it just not, you know, it doesn't

42:09

break in a nice way. It just, you know, we'll

42:11

go halfway through a question and then say, I am unable

42:14

to produce a response.

42:16

I really love the idea that there's some incredibly

42:18

well played, you know, some of the smartest

42:21

kind of machine learning engineers,

42:24

you know, sat next to some of the world's,

42:26

brightest trust and safety experts, some

42:28

of whom we know probably, um, you know,

42:30

sat together being like, what are the list of six

42:32

names that we've gotta add to a block list?

42:35

Yeah. And it's just, it's so obviously

42:37

just a straight up lock list that says we

42:39

will not produce these names and people have gone

42:41

through and reasoned out some of them and they're

42:44

all slightly different. There's a few

42:46

that we don't quite know why

42:48

there was one person, Guido Scorza,

42:50

who is a data protection

42:53

expert in Italy, and he had

42:55

posted that he had used the

42:57

GDPRs. right to erasure

43:00

process against open AI

43:02

and said, basically you have to forget everything about

43:04

me. And the only way that they could figure

43:07

out to do so was to put him on this

43:09

block list Because

43:11

they don't actually remember anything about him.

43:13

Like there's no this is the thing that a lot of people

43:15

don't Understand about these systems

43:17

is it's not a big database. It's not going in

43:19

and collecting stuff. It is training and

43:21

just thinking about stuff. So the only way to say like,

43:24

never produce any information

43:26

on Guido Scorza is to put

43:28

them on a, I mean, there are better ways,

43:31

but this was the sort of fast and quick way.

43:33

There are other people, you know, the, Brian

43:35

Hood was a mayor in

43:37

Australia who claimed that

43:39

it produced defamatory content about him

43:42

and he threatened to sue. And so. that

43:44

one probably came up first. Somebody had actually

43:46

told me about that one, like a year and a half ago when

43:48

it, when it came up and I had meant to investigate it. And then

43:50

I just never got to it. Jonathan Turley,

43:53

who's a law professor and sort

43:55

of a famously sort of Trumpian

43:57

law professor. Went on this big

44:00

rant of claiming also that chat

44:02

GPT had defamed him. And so he's on the

44:05

The weird one is Jonathan Zatrain, who

44:07

I know, and asked

44:09

like, what the hell, what

44:11

yeah,

44:12

to get off this list? And he has no

44:14

idea.

44:15

right, right.

44:17

he doesn't know what's going on. I joked

44:19

with him that like, Oh, you know, are you trying to

44:21

come up with something to prove a point or something

44:24

about all this? He's like, I did nothing.

44:28

So,

44:28

And he didn't know he was on ChachiPT's

44:31

naughty step.

44:31

he, he had found it, he had actually discovered

44:33

it. So he, he added himself when everyone

44:35

was pointing out to this list of names, he pointed out,

44:37

like he had discovered a few months ago that

44:40

chat GPT breaks on his name. and

44:42

so, you know, it's just. Basically,

44:45

it's like this really simple way. I understand

44:47

probably why it happened because all

44:49

these people, except for Jonathan, as far as we know,

44:52

got really mad. We still don't know which David

44:54

Mayer there is. There's speculation. There's a few different,

44:56

there's also a David Farber, Faber, and

44:58

nobody's sure which one that is. but

45:00

the other ones, like people got mad and open

45:03

AI probably was like, you know what, this

45:05

guy is a pain. Fucking headache. And

45:07

therefore just put them on the ban list.

45:10

Like, let's not deal with this because we don't want to

45:12

go through a lawsuit and have to do

45:14

this. And there's no other way to like

45:16

effectively, because they're, you know, they don't understand.

45:19

It's not that chat GPT is defamatory.

45:21

It's whatever the prompt is led to defamation.

45:25

And it's not that we're collecting data on

45:27

this guy, but you know, he thinks

45:29

we are, and do we really want to fight it in

45:31

court? I just put them on the ban

45:32

Put him on the list.

45:33

I'm pretty sure that's the kind of thing that

45:35

happened. You know, you would hope that

45:37

a company with as many resources, and as

45:39

you noted, as many, smart

45:42

engineers would come up with a more sophisticated

45:44

way than this, but they haven't. And so,

45:47

you know, I know how things like that happen and

45:49

I'm sure it was just kind of like, ah, just make this

45:51

headache go away in the easiest way possible. And

45:53

that means every time you try and produce this name,

45:56

we're going to break it.

45:57

Yeah. Interesting. just as an

45:59

aside, there's some, there's the OpenAI put out

46:01

some research in the past week around

46:03

kind of red teaming in new, new and novel ways.

46:06

I wonder if they should be reading their own research.

46:08

Yeah, yeah, absolutely,

46:11

absolutely.

46:12

so yeah, if you, if David May is listening,

46:14

to the podcast, if he's, if he's a regular listener,

46:16

um, make yourself known, um,

46:19

we would love to know what you did to be put on

46:21

the, uh, on the shit list of, uh,

46:23

Yeah. And then you got taken off. So if

46:25

you have a legal complaint, get it, get it going

46:28

Yeah, we know some lawyers.

46:30

Um, great. Thanks Mike. Um,

46:33

that's not the, you know, that's not the kind of end

46:35

of today's episode. It's, we've got a great bonus

46:37

chat. I don't know if you want to kind of give us a preview

46:39

of that.

46:40

Yeah. So this was really great. I had this discussion

46:42

just yesterday. it's a topic that we've talked about

46:44

a lot. The social media age bands

46:47

in Australia and how they're popping

46:49

up elsewhere. and so this is a discussion

46:51

I had with, two folks from the internet society,

46:53

Natalie Campbell, who's the senior director

46:56

of North American government and regulatory affairs.

46:58

And John Perrino, who's a senior

47:00

policy and advocacy expert. And

47:03

they're concerned about these

47:05

age restrictions and age verification.

47:08

And, we talked about Australia

47:10

and also Canada has a bill that

47:13

is, moving forward that one of the incredible

47:15

things in the chat was that,

47:17

it would require places like Starbucks

47:20

to verify the age of anyone who wants

47:22

to use the wifi in Starbucks

47:25

and, you know, all sorts of stuff. And

47:27

so they, they're sort of concerned about.

47:29

these laws, the proliferation of these laws,

47:31

what it means for the open internet and

47:34

what age verification requirements

47:36

would mean. The really fascinating discussion

47:38

and we'll go to that right now. Natalie,

47:51

John, welcome to the podcast. Glad to have you

47:53

here. wanted to start by talking about

47:55

the law that just passed in Australia, which

47:57

effectively bans those under 16

48:00

from social media. John, can you

48:02

talk through the details of that law, including

48:04

sort of which sites are impacted and

48:07

how do you, Does the Australian government

48:09

expect sites to know the age of

48:11

their visitors?

48:14

great question, Mike, and,

48:16

you know, for, for quick background for

48:18

listeners, this legislation

48:21

moved through in about less

48:23

than two weeks and

48:25

there was a consultation period open less than

48:28

24 hours. Um,

48:31

the Australian social media age verification

48:33

stuff really, really flew through. And

48:36

there's a lot that we still don't know, even

48:38

though the legislation passed, especially

48:40

on what social

48:42

media platforms would be

48:45

required to do in order to

48:47

comply with this. So

48:49

right now, essentially all we know

48:52

is that government ID

48:54

would not be required. That should

48:56

include things like a passport or government

48:59

issued driver's license, also

49:01

a digital ID. And

49:03

we know that the social media is

49:06

for Australians under

49:08

the age of 16. and

49:10

then the final thing that was kind of added on late

49:13

was a digital duty of

49:15

care, which again is to

49:17

be determined. So the bottom

49:19

line on this is much

49:21

of the bill is to be determined. They

49:23

don't know what age

49:25

verification would be,

49:28

as they say, reasonable

49:30

for the social media platforms to use.

49:32

And as a lot of the comments pointed out,

49:35

there was actually already an age verification

49:38

study ongoing, and

49:41

that's still going on, which tools,

49:43

maybe work best, what the trade offs

49:45

are in different age verification methods.

49:48

The Australian government was already doing this. They

49:50

don't have the results. They probably won't

49:52

have. The results until about

49:55

six months before the social

49:57

media platforms have to comply with this.

50:00

I know that we've seen a bunch of other countries also exploring

50:03

similar ideas around age verification.

50:06

on the podcast, we've talked about it in

50:08

the UK and somewhat in

50:10

the US to a lesser extent, there are a couple of different

50:12

issues in the US, but also

50:14

now Canada is exploring a

50:16

similar issue. So, Natalie,

50:18

can you talk through what the proposal

50:21

in Canada is about?

50:22

Sure, so, Senate Bill S 210,

50:25

this is an act to restrict young persons online

50:28

access to sexually explicit material,

50:30

is very close to becoming a law.

50:33

This bill has been flying under the radar

50:35

for quite a long time. I don't think

50:37

that people are taking it very seriously because

50:40

most Senate bills. Don't make

50:42

it to law. but this has found

50:44

its way into the very late stages

50:47

of our parliamentary process,

50:49

and it's probably the most dangerous

50:52

bill to the Internet, in

50:54

Canada right now. and one

50:56

of the main reasons for that is because of

50:58

its age verification mandates.

51:00

That would apply to virtually every

51:02

intermediary on the internet and

51:05

some that are not on the internet as well.

51:08

So essentially what the bill tries to do is

51:10

to prevent young people from access

51:12

to sexually explicit material, which

51:14

is defined extremely broadly.

51:17

and makes it so that every intermediary

51:20

that would play a part in facilitating

51:23

sexually explicit material for profit online

51:26

would have to verify a user's

51:28

age or face very high fines.

51:31

So, this, you know, is this,

51:33

is this, Probably the most extreme

51:35

age verification proposal we've seen so far,

51:37

because it's not just targeting websites,

51:40

it's targeting internet service

51:42

providers, content delivery networks,

51:45

search engines, email services,

51:47

even a Starbucks location that's providing access

51:50

to Wi Fi. Because

51:52

what content is flowing through this pipes

51:54

could be sexually

51:56

explicit material. they now have

51:59

a duty to verify

52:01

users age and in doing

52:03

so like it's, it's different. When

52:05

we're talking about websites doing

52:08

this, but when we're thinking about every single

52:10

intermediary on the Internet having a duty

52:12

to verify users age, that

52:14

gets really problematic when we're thinking

52:16

about an open Internet first,

52:19

because most traffic

52:21

is encrypted, and it's not

52:23

possible for most infrastructure

52:26

intermediaries to know what's flowing

52:28

through the pipes, and even if they could, content

52:30

doesn't flow through the Internet as a

52:32

whole piece of content. It's packets.

52:35

And so it becomes extremely difficult

52:37

to, one, identify, what

52:39

is sexually explicit material and,

52:42

you know, is Canada's definition

52:44

of this type of content. And

52:46

then, second of all, having

52:49

to, decide to try

52:51

and identify that stuff means you can't use things like

52:53

encryption, which is the foundation

52:55

of security for every service

52:58

and user on the Internet. So

53:00

it's very concerning in that, the implications

53:03

for security online, but also

53:05

the fact that you're placing

53:07

huge barriers to access because now

53:09

is an Internet user, because

53:11

a lot of intermediaries. Won't know

53:14

what's sexually explicit material. They

53:16

might just start doing age verification

53:19

to Everyone and that means

53:21

I now, you know as somebody

53:23

who's based in canada would have to trust

53:25

a whole lot of entities

53:28

with Very personal information

53:30

whether it's government id or

53:33

you know biometrics i'm having

53:35

to trust a lot of third

53:37

parties Who I might not have?

53:40

You Any direct relationship with with

53:42

my personal information, which is a huge

53:45

barrier to privacy and anonymity

53:47

online. so this is, from

53:49

an open Internet aspect, this is super

53:52

problematic because you're creating huge barriers

53:54

to access. for people who

53:56

might not be able to get government issued I. D.

53:58

or might not be able to use the Internet without the

54:00

promise of anonymity, which

54:02

is a huge hurdle for a lot of people in marginalized

54:05

communities and young people as well. But

54:07

also the fact that this has huge implications

54:10

for security online. and

54:13

you know, not Enabling

54:15

intermediaries to use encryption

54:18

that could be extremely devastating to

54:20

people's security online and making people

54:23

vulnerable to a whole range of bad

54:25

stuff that I don't think was the intention

54:27

of this bill.

54:28

Yeah, I mean you, you mentioned just really

54:30

quickly. I wanted to follow up on one point. You mentioned

54:32

the fact that since it hits every intermediary,

54:35

the potential that, like a Starbucks,

54:38

if you wanted internet in a Starbucks, is

54:40

the idea there that, Starbucks would have

54:42

to not just check your ID, but like

54:44

record it somehow? Is that part of the fear

54:46

that if you want to get on the Wi Fi at Starbucks,

54:49

you have to first prove how old you are?

54:51

Well, this is the problem

54:53

is that the way that it defines

54:56

who this bill applies to is

54:58

internet service provider, but the definition

55:01

it uses for that is a

55:03

person who provides access to the Internet

55:06

Internet content hosting or electronic

55:09

mail. The person who provides

55:11

internet access could mean

55:13

Starbucks and could mean that, you

55:16

Starbucks has to know what are you

55:18

looking at when you're accessing the internet or

55:21

just say, just show us your ID if you want to use

55:23

the Wi Fi, right? So, problematic

55:26

in both senses and just a reminder

55:28

that please use VPNs. Okay.

55:32

So, I mean, so we're seeing this all

55:34

over, obviously, Australia and Canada,

55:36

as you guys discussed, we've discussed the UK in the

55:38

past. There are various bills in the U S that

55:40

also touch on this elsewhere around the world.

55:42

What should we make of this trend? What,

55:44

you know, what, why is this all, all of a sudden

55:46

happening? We've had the internet obviously for decades

55:49

now. Why is it suddenly that

55:52

everyone is trying to pass these kind of

55:54

age related bans or age verification.

55:57

ideas, John, do you want to start

55:59

there?

56:00

Yeah, I mean, to that point, it certainly feels

56:02

like everything is happening

56:05

all at once everywhere

56:07

in every corner of the world. I

56:09

mean, we're seeing this, as you mentioned, you

56:11

know, the UK has been working on this almost

56:14

a decade, but their age verification

56:16

guidance comes out in January. The

56:19

European Union is working on this. There

56:21

are almost two dozen

56:24

U. S. states that have passed age

56:26

verification for adult websites,

56:29

more with social media. it

56:31

really is happening everywhere. Why?

56:33

You know, it's a whole bunch of factors.

56:35

Um, one kind of interesting story

56:38

with the Australia legislation

56:40

is it seems to have stemmed depending

56:43

on who you ask, from Jonathan Haidt's

56:45

book. and, you know, there's,

56:47

has been a lot of discussion,

56:50

because of recent literature.

56:53

The Anxious Generation is Jonathan Height's

56:55

book, as I'm sure many of your listeners are

56:57

familiar with. There's just been more

56:59

discussion on this. There's been a lot of discussion

57:02

in the U. S. You know, the Kids Online Safety

57:04

Act, but there are, you know, true age verification

57:07

bills being introduced. There's

57:09

even a case going to the Supreme Court.

57:11

And that's the Free Speech Coalition v. Paxton

57:13

case. And that's something that the Internet Society

57:16

Weight in on with an amicus. We joined with the

57:18

Center for Democracy and Technology, New

57:20

America's Open Technology Institute, and

57:23

some academics on that case. And,

57:25

you know, the thing that we think doesn't get talked

57:27

about enough is that age verification

57:30

laws are not just about

57:32

young people.

57:33

to

57:34

verification laws are about everyone,

57:37

if not done right. You

57:39

know, this puts everyone's privacy

57:41

security at risk and

57:43

like Natalie already said, you know, this can be discriminatory

57:46

this the marginalized communities

57:48

and there's really interesting research on

57:50

this marginalized communities can benefit

57:52

the most from social media can benefit

57:54

the most from being online. It

57:56

can be more difficult for them to get online. There

57:59

can be more social factors that make it.

58:01

Um,

58:05

you know, generally their worlds, it's

58:07

more difficult. To make those types of connections.

58:10

so what we're really focused on

58:13

is that everyone can get

58:15

access to the Internet that is not. Gating

58:17

off access to the Internet,

58:19

to news, information, health, entertainment,

58:23

right? so this is, this is really

58:25

going to be a challenge and we see so

58:27

many pieces of regulation that are being

58:29

implemented and introduced right

58:31

now. luckily at the Internet Society,

58:34

we have an incredible community. for

58:36

instance, our Australia chapter.

58:38

Jumps to action. There was less than 24

58:40

hours to file, comments,

58:43

they made the deadline, got the comments

58:45

in, you know, and that's so important.

58:48

And so many of our, our local,

58:50

you know, regional chapters are engaging

58:53

with governments in

58:55

their home countries on this issue. So

58:58

on that point, you know, there's really good

59:00

debate on this. really great interaction.

59:04

More and more technologists are getting involved

59:06

in this, more standards development organizations.

59:09

So the engineers who actually make the internet

59:11

function, right, are getting involved

59:13

on this. So that's encouraging, but it really

59:15

is happening everywhere at once. And

59:18

we really need to make sure that those

59:21

who need the internet the most, can actually

59:23

be safe online and are not having their

59:25

privacy and security exposed through all

59:27

this.

59:29

you've both talked a little bit about sort of the

59:31

implications of this and the reasons

59:33

why there are concerns, but

59:36

Natalie, I want to finish up with you and just

59:38

say, you know, for policymakers, some

59:40

of whom might be listening to this, hopefully, um,

59:42

who are looking at these laws and thinking

59:45

about them, what kinds of.

59:47

Factors, what should they be considering before,

59:50

you know, proposing or voting

59:52

for these kinds of laws?

59:54

So, first, like, it's really

59:56

important to understand that the Internet Society works

59:58

to make sure that the Internet is for everyone.

1:00:01

And when we talk about an open

1:00:03

Internet, we're talking about making

1:00:06

sure we're lowering barriers to access to the Internet.

1:00:08

We also want a healthy Internet.

1:00:11

We are, you know, like John

1:00:13

mentioned, we're a huge community, not just the Internet

1:00:15

Society, but our chapters and members and organizational

1:00:17

members around the world. We all want

1:00:19

a healthy Internet and care about making

1:00:22

sure there are safe spaces for people online. But

1:00:24

we also want to make sure that, you know, things

1:00:27

like encryption and

1:00:29

the fundamentals of a secure Internet.

1:00:31

Are not undermined and that people

1:00:34

don't experience barriers to access,

1:00:37

that could be complete hurdles

1:00:39

to accessing the Internet in the first place. So,

1:00:42

I mean, it's not to say that there might

1:00:44

never be a solution for age verification

1:00:47

that could not hinder things like

1:00:49

security and open Internet. But

1:00:52

policymakers really do have to be thinking

1:00:54

through how could their proposals.

1:00:57

Impact, people's access to the Internet

1:00:59

and their safety online. And

1:01:01

we have a tool that helps us analyze

1:01:04

these proposals is called the Internet Impact Assessment

1:01:06

Toolkit. We think of it like an environmental

1:01:09

impact assessment for the Internet.

1:01:11

And so what we do, is, um,

1:01:14

Offer ourselves to work with governments

1:01:16

who are thinking about, whatever

1:01:18

issues are working out that may relate to the Internet.

1:01:21

And we'll often use this framework

1:01:23

that describes what the Internet needs to exist

1:01:26

in the 1st place and to be more open,

1:01:28

global connected, secure and trustworthy and

1:01:30

we help them think through how might. Particular

1:01:33

proposal impact these goals

1:01:35

for a healthy Internet. And

1:01:37

so always are available

1:01:39

to work with governments to think through these aspects,

1:01:42

but we have a toolkit that policymakers

1:01:44

can use themselves. And we

1:01:46

think that it's really important that,

1:01:48

We just don't jump to law

1:01:50

proposals that don't consider those impacts

1:01:52

on the internet because such as

1:01:54

a case in Canada, there can be

1:01:57

very extreme consequences for

1:01:59

people's access to the internet and their

1:02:01

very safety and security online.

1:02:03

All right. Well, Natalie and John,

1:02:05

thank you for coming on the podcast and thank

1:02:08

you for all the good work that the Internet Society

1:02:10

does. And, uh, I hope if

1:02:12

anyone's listening to this and you are, working

1:02:15

on a bill like this or thinking about these kinds

1:02:17

of laws that you, uh, listen closely

1:02:19

and, take a look at what the Internet Society is

1:02:22

doing on this and, and what, resources

1:02:24

they have available. Thanks again

1:02:26

for joining us.

1:02:27

Thanks Mike.

1:02:27

Thanks, Mike.

1:02:31

Thanks for listening to Ctrl-Alt-Speech.

1:02:34

Subscribe now to get our weekly episodes

1:02:36

as soon as they're released. If your

1:02:38

company or organization is interested in sponsoring

1:02:40

the podcast, contact us by visiting

1:02:43

ctrlaltspeech.Com. That's

1:02:45

C T R L Alt Speech. com.

1:02:48

This podcast is produced with financial support

1:02:50

from the Future of Online Trust and Safety Fund,

1:02:53

a fiscally sponsored multi donor fund

1:02:55

at Global Impact that supports charitable

1:02:57

activities to build a more robust, capable,

1:02:59

and inclusive trust and safety ecosystem.