0:04

Welcome to the Tech Meme Right Home for Tuesday, September 24th,

0:06

2024. I'm Brian

0:08

McCullough today. Sam Altman has a

0:10

manifesto around AI, kind of.

0:13

Telegram is beginning to walk things back a

0:15

bit. Cloudflare wants to help you block the

0:18

AI bots. New streaming device from Roku, and

0:20

I guess TikTok can't win at everything. Here's

0:22

what you missed today in the world of

0:24

tech. Sam

0:33

Altman released an essay yesterday where

0:35

he said that superintelligence is right

0:37

around the corner. Quote,

0:40

this may turn out to be the

0:42

most consequential fact about all of history

0:44

so far. It is possible that we

0:46

will have superintelligence in a few thousand

0:48

days. It may take longer,

0:50

but I'm confident we'll get there. How did we

0:53

get to the doorstep of the next leap in

0:55

prosperity? In three words, deep

0:57

learning worked. In 15 words,

0:59

deep learning worked, got predictably better

1:01

with scale, and we dedicated increasing

1:03

resources to it. That's really

1:06

it. Humanity discovered an algorithm that could

1:08

really, truly learn any distribution of data,

1:10

or really the underlying rules that produce

1:12

any distribution of data, to a shocking

1:14

degree of precision. The more compute and

1:17

data available, the better it gets at

1:19

helping people solve hard problems. I

1:21

find that no matter how much time I spend

1:23

thinking about this, I can never really internalize how

1:26

consequential that is. There

1:28

are a lot of details we still have to

1:30

figure out, but it's a mistake to get distracted

1:32

by any particular challenge. Deep

1:34

learning works, and we will solve the remaining

1:36

problems. We can say a lot of things

1:38

about what may happen next, but the main

1:40

one is that AI is going to get

1:42

better with scale, and that will lead to

1:44

meaningful improvements to the lives of people around

1:47

the world. End quote. Now,

1:49

I will be honest with you. Despite

1:52

all the attention this got online overnight, the

1:54

whole thing is sort of vague and

1:57

extremely cheer-lady kind of. like,

2:00

is this just some hype cycle chum?

2:02

I mean, probably, right? To a degree.

2:04

Though now we do have this clam

2:06

chowder, as Gruber used to call it,

2:08

to bring up and be like,

2:11

hey Sam, it's been a thousand days. Quoting

2:14

TechCrunch. In Altman's latest

2:16

post, the OpenAI leader presents an incredibly positive

2:18

update on the state of AI, hyping its

2:20

world-changing potential. Far from being

2:22

an occasionally helpful alternative to Google Search,

2:24

or a homework helper AI, as Altman

2:27

presents, will change humanity's progress for the

2:29

better, naturally. Through rose-tinted contacts,

2:31

Altman pitches the numerous ways he

2:33

believes AI will save the world.

2:36

But much of what he writes is seemingly

2:38

meant to convince the skeptics of how much

2:40

AI matters and could well have the opposite

2:42

result. Instead of creating new fans, posts like

2:44

this may well invite increased

2:46

scrutiny as to whether we're in an

2:48

Emperor's New Clothes situation. As

2:51

one commentator with the username Shark Jacobs on

2:53

the technical forum Hacker News writes, I'm not

2:55

an AI skeptic at all. I use LLMs

2:57

all the time and find them very useful,

3:00

but stuff like this makes me very skeptical

3:02

of the people who are making and selling

3:04

AI." Then

3:07

quoting from Ars Technica. The

3:09

OpenAI chief envisions AI assistants becoming

3:11

increasingly capable, eventually forming quote, personal

3:14

AI teams that can help individuals

3:16

accomplish almost anything they can imagine.

3:18

He predicts AI will enable breakthroughs in

3:21

education, healthcare, software development, and other fields.

3:23

While acknowledging potential downsides and labor

3:26

market disruptions, Altman remains optimistic about

3:28

AI's overall impact on society. He

3:30

writes quote, "...prosperity alone doesn't

3:32

necessarily make people happy. There are plenty

3:34

of miserable rich people, but it would

3:36

meaningfully improve the lives of people around

3:38

the world." End quote. Even with AI

3:41

regulation like SB 1047, the

3:43

hot topic of the day, Altman

3:45

didn't mention sci-fi dangers from AI

3:47

in particular. On X,

3:49

Bloomberg columnist Matthew Iglesias wrote, "...notable

3:52

that at Sama is no longer even

3:54

paying lip service to existential risk concerns,

3:57

the only downsides he's contemplating are labor market

3:59

adorables." adjustment issues." While

4:02

enthusiastic about AI's potential, Altman urges caution

4:04

too, but vaguely. He writes,

4:06

we need to act wisely but with

4:08

conviction. The dawn of the intelligence age

4:11

is a momentous development with very complex

4:13

and extremely high stakes challenges. It

4:15

will not be an entirely positive story, but the upside

4:17

is so tremendous that we owe it to ourselves and

4:19

the future to figure out how to navigate the risks

4:21

in front of us. US

4:29

customers of Kaspersky antivirus, guess

4:31

what? You have a

4:34

new antivirus program whether you know

4:36

it or not. That's because Ultra

4:38

AV was installed remotely to replace

4:40

Kaspersky on your PCs after a

4:42

Kaspersky deal with Ultra AV owner

4:45

Pango. Quoting TechCrunch, customers

4:47

of Kaspersky antivirus in the United States

4:49

found out in the last few days

4:51

that their cybersecurity software was automatically replaced

4:54

with a new one called Ultra AV according

4:56

to several customers who posted on social media.

4:59

Some of the customers said they had no idea this was going

5:01

to happen. Quote, woke up

5:03

to Kaspersky completely gone from my system

5:06

with Ultra AV and Ultra VPN freshly

5:08

installed, not by me, just automatically while

5:10

I slept, a user on Reddit

5:12

wrote. Others in the same Reddit thread

5:14

as well as in other threads reported having the

5:17

same experience. A reseller who

5:19

until recently sold Kaspersky products prior to

5:21

the recent salesman told TechCrunch that he

5:23

was left annoyed by the move to

5:25

automatically remove Kaspersky software and replace it

5:27

with an entirely different antivirus. A

5:30

former senior US government cybersecurity official said

5:32

that this was an example of the

5:34

huge risk posed by the access granted

5:36

by Kaspersky software. It's worth

5:38

noting that on the other hand, other customers

5:41

did report receiving an email from Kaspersky about

5:43

the transition to Ultra AV. The

5:45

transition to Ultra AV comes months after the

5:48

US government took the unprecedented decision to ban

5:50

all sales of Kaspersky software across the United

5:52

States. In June, the Commerce Department

5:54

announced that sales of the antivirus software would

5:56

be banned starting July 20th. to

6:00

provide limited security updates to customers until

6:02

September 29th. In early September,

6:04

Axios reported that Kaspersky had reached

6:06

a deal to offload its customers

6:08

to American cybersecurity firm Pango, which

6:11

owns UltraAV, a relatively new antivirus

6:13

software." Telegram

6:20

now says it will give users IP

6:22

addresses and phone numbers to authorities in

6:24

response to valid legal requests, changing

6:27

its terms of service to deter criminal

6:29

activity on the platform. Quoting

6:31

Bloomberg, The platform changed its terms of

6:33

service to deter criminals from abusing it,

6:35

Durov said in a post on Telegram

6:37

Monday. The move comes less than a

6:39

month after his arrest in France, where

6:41

he faces charges of alleged complicity in

6:43

the spread of child sexual abuse materials.

6:46

The move represents a marked difference from

6:48

Telegram's approach to government requests for data

6:50

and its reputation for lax moderation. The

6:53

United Arab Emirates-based platform has been notoriously

6:55

non-responsive to take down requests from governments

6:57

around the world and often ignored requests

6:59

for information about suspected criminals. The

7:02

app, using artificial intelligence and a team

7:04

of moderators, has now begun to conceal

7:06

problematic content from its search results as

7:09

part of its efforts to prevent misuse,

7:11

Durov said. French prosecutors in

7:13

August charged the Russian-born Durov in connection with

7:15

alleged crimes committed on the app. The case

7:17

portrays the billionaire CEO as the head of

7:20

a company that refused to provide law enforcement

7:22

with data to assist legal wiretaps on suspected

7:24

criminals. Durov's arrest has led

7:26

to a number of policy changes at

7:28

Telegram. Earlier this month, Telegram disabled new

7:30

media uploads, which Durov said was aimed

7:32

at stopping bots and scammers." CloudFlare

7:41

has rolled out bot management, a

7:43

suite of free AI auditing tools

7:45

meant to help monitor and selectively

7:47

block AI-scraping bots to

7:50

all of its customers. Quoting Wired, crawl

8:00

for data. Today it's giving all of

8:02

its customers, including the estimated 33 million using

8:04

its free services, the ability to monitor

8:06

and selectively block AI data scraping bots.

8:09

That preventative measure comes in the form of

8:11

a suite of free AI auditing tools it

8:13

calls bot management, the first of which allows

8:15

real-time bot monitoring. Customers will have

8:17

access to a dashboard showing which AI crawlers

8:20

are visiting their websites and scraping data, including

8:22

those attempting to camouflage their behavior. We've

8:25

labeled all the AI crawlers, even if

8:27

they try to hide their identity, says

8:29

CloudFlare co-founder and CEO Matthew Prince, who

8:31

spoke to Wired from the company's European

8:33

headquarters in Lisbon, Portugal, where he's been

8:35

based the past few months. CloudFlare

8:37

has also rolled out an expanded bot

8:39

blocking service, which gives customers the option

8:41

to block all known AI agents or

8:43

block some and allow others. Earlier

8:46

this year, CloudFlare debuted a tool that allowed

8:48

customers to block all known AI bots in

8:50

one go. This new version offers more control

8:52

to pick and choose which bots they want

8:54

to block or permit. It's a

8:56

chisel rather than a sledgehammer, increasingly useful as

8:58

publishers and platforms strike deals with AI companies

9:01

that allow bots to roam free. We want

9:03

to make it easy for anyone, regardless of

9:05

their budget or their level of technical sophistication,

9:07

to have control over how AI bots use

9:10

their content, Prince says. CloudFlare

9:12

labels bots according to their functions,

9:14

so AI agents used to scrape

9:16

training data are distinguished from AI

9:18

agents pulling data for newer search

9:20

products like OpenAI's search GPT. Websites

9:23

typically try to control how AI bots

9:26

crawl their data by updating a text

9:28

file called robots exclusion protocol or robots.txt.

9:30

This file has governed how bots scrape

9:32

the web for decades. It's not illegal

9:34

to ignore robots.txt, but before the age

9:36

of AI, it was generally considered part

9:39

of the web's social code to honor

9:41

the instructions in the file. Since the

9:43

influx of AI scraping agents, many

9:45

websites have attempted to curtail unwanted

9:48

crawling by editing their robots.txt files.

9:50

Services like the AI agent watchdog DarkVisitors

9:52

offer tools to help website owners stay

9:55

on top of the ever increasing number

9:57

of crawlers they might want to block,

9:59

but they've been limited by a

10:01

major loophole. Unscrupulous companies tend to

10:03

simply ignore or evade robots.txt commands.

10:06

According to Dark Visitors founder Gavin King,

10:08

most of the major AI agents still

10:10

abide by robots.txt. That's been

10:12

pretty consistent, he says. But not all

10:14

website owners have the time or knowledge

10:16

to constantly update their robots.txt files, and

10:19

even when they do, some bots will

10:21

skirt the files' directives. They try to

10:23

disguise the traffic. Prince says

10:25

Cloudflare's bot-blocking won't be a command that

10:27

this kind of bad actor can ignore.

10:30

Robots.txt is like putting up a no trespassing

10:32

sign, he says. This is like having a

10:34

physical wall patrolled by armed guards. Just

10:37

as it flags other types of suspicious

10:39

web behavior like price-scraping bots used for

10:41

illegal price monitoring, the company has created

10:43

processes to spot even the most carefully

10:45

concealed AI crawlers, end quote. Lumen

10:54

is the world's first handheld metabolic coach.

10:56

It's a device that measures your metabolism

10:58

through your breath, and on the

11:00

app, it lets you know if you're burning fat

11:02

or carbs and gives you tailored guidance to improve

11:04

your nutrition, your workouts, sleep, even stress management. All

11:06

you have to do is breathe into your lumen

11:09

first thing in the morning and you'll know what's

11:11

going on with your metabolism, whether you're burning mostly

11:13

fats or carbs. Then Lumen gives

11:15

you a personalized nutrition plan for that

11:18

day based on your measurements. I've been

11:20

using Lumen, I used it this morning.

11:22

You can also breathe into it before and

11:24

after workouts and meals, which I found really

11:26

useful so you know exactly what's going on in your body in

11:28

real time, and Lumen will give you tips to keep you on

11:31

top of your health game. It can

11:33

also track your cycle as well as

11:35

the onset of menopause and adjust your

11:37

recommendations to keep your metabolism healthy through

11:39

hormonal shifts, so you can keep up

11:41

your energy and stave off cravings. So

11:43

if you want to take the next

11:45

step in improving your health, go to

11:47

lumen.me slash ride to get 15% off

11:49

your Lumen. That's L-U-M-E-N dot

11:52

M-E slash ride for 15% off

11:54

your purchase. Thank you, Lumen, for sponsoring this

11:56

episode. Man,

12:01

when we leave the house, it's phone wallet

12:03

keys, how's my hair look, right? But if

12:05

you're experiencing hair loss, you may not be

12:07

so confident when you step outside your door.

12:10

So it's time to get that confidence back and

12:12

restore your hair with HIMS. HIMS

12:14

provides access to a range of hair loss treatments that

12:16

work, all from the comfort of your couch. HIMS

12:19

makes treating hair loss simple with doctor-trusted

12:21

treatment options and clinically proven ingredients like

12:24

finasteride and minoxidil that can regrow hair in

12:26

as little as three to six months. For

12:28

personalized, chewable, oral spray and serum treatment options

12:30

so you can find what works best for

12:33

you. The process is simple and 100% online.

12:36

Just answer a few questions and a medical provider

12:38

will determine if treatment is right for you. If

12:40

prescribed, your treatment is sent directly to your door.

12:43

Start your free online visit

12:46

today at hims.com/Ride. That's hims.com/Ride

12:48

for your personalized hair loss

12:51

treatment options. hims.com/Ride. Results

12:53

vary based on studies of topical

12:55

and oral minoxidil prescription products, requests,

12:57

and online consultation with a healthcare

12:59

provider who will determine if a

13:02

prescription is appropriate. Restrictions apply, see

13:04

website for full details and important

13:06

safety information. Roku

13:10

has unveiled the $100 Roku Ultra,

13:12

which they say is 30% faster

13:14

than other streaming players,

13:16

has Wi-Fi 6, HDMI

13:19

2.1, Dolby Atmos, and the Voice

13:21

Remote Pro and more. Quoting

13:24

the Verge. One day

13:26

after the Google TV streamer hit stores,

13:28

Roku is announcing its latest streaming player

13:30

and unlike Google's big design overhaul, the

13:32

new Roku Ultra is playing things safe.

13:35

I'd argue maybe too safe. It's

13:37

still $99.99, still delivers Dolby

13:39

Vision and Atmos, and it looks exactly the same

13:41

as before, though the company says

13:43

there are notable improvements inside. The

13:46

2024 Ultra is 30% faster than any

13:48

of Roku's other players and it now features Wi-Fi

13:50

6. That upgrade

13:52

puts it ahead of the Google TV streamer,

13:54

which stuck with Wi-Fi 5. Roku

13:57

is now on par with the Apple

13:59

TV, but behind Amazon's flagship Fire TV

14:01

streamers, which include Wi-Fi 6E connectivity. And

14:03

yes, there's still a built-in ethernet jack

14:05

for those of you who prefer wired

14:08

setups. Plus, as an

14:10

HDMI 2.1 streaming player, Roku

14:12

Ultra offers quick media switching, QMS,

14:15

delivering seamless, judder-free transitions between different

14:17

refresh rates or content sources on

14:19

premium TVs. Roku's Paul

14:21

Nagaroni wrote in a blog post, note that

14:24

your TV must also support QMS for this

14:26

to work. And we're still in the

14:28

very, very early stages of that. The

14:30

Ultra ships with Roku's best remote, the

14:32

Voice Remote Pro, which itself has been

14:34

upgraded with backlit buttons, USB-C charging, and

14:36

new quick launch and live TV buttons.

14:39

The former lets you, quote, see your personalized

14:41

shortcuts, your go-to apps, your favorite voice command,

14:43

switching the captions on and off right on

14:45

your TV screen. The

14:47

2024 Roku Ultra also, quote, uses advanced machine

14:50

learning to anticipate which app you are going

14:52

to open next and tease it up for

14:54

our snappiest app loading to date, end quote.

14:57

Spotify has rolled out its AI playlist

15:00

feature in beta to premium subscribers in

15:02

the US, Canada, Ireland, and New Zealand

15:04

after launching in the UK and Australia,

15:10

quoting The Verge. Subscribers

15:13

can locate the feature within the mobile app by

15:15

tapping the plus button at the top right of

15:18

their Spotify library. Selecting AI

15:20

playlist from the dropdown menu will then open a

15:22

chat box to describe the

15:24

playlist you want, such as spooky songs

15:26

to play during Halloween or something folk

15:28

music for a chill bath time. The

15:31

feature will also provide suggested prompts. The

15:33

AI playlist beta isn't currently available on

15:36

the Spotify desktop or web apps. The

15:39

generated playlists contain 30 songs that can

15:41

be adjusted with additional prompts to better

15:43

match the desired vibe, such as asking

15:45

for more upbeat music, descriptions that specify

15:47

things like genres, decades, moods, or artists

15:50

will see better results. But

15:52

in my own testing, I found it

15:54

was impressively capable of mashing songs to

15:56

even niche descriptions like, make me feel

15:58

like a vampire hunter from Blade 1998."

16:00

And finally, TikTok says it plans to

16:02

shut down

16:11

TikTok Music on November

16:14

28th will be deleting all customer data.

16:16

Did you even know TikTok had a

16:18

music app? Exactly. Quoting

16:21

Bloomberg, Bite Dance TikTok's owner

16:23

created the music service to capitalize on the

16:26

app's power and popularity with music fans. TikTok

16:28

videos have been credited with the rise of

16:30

hit songs and new artists and have led

16:32

to clashes with record labels which have demanded

16:34

higher payments for using their clips. The

16:37

music app, originally called Resso, was launched initially in

16:39

2019 in a bet that the company

16:42

could funnel TikTok users into a

16:44

new standalone music service. It was

16:46

available in Brazil, Indonesia, Australia, Mexico,

16:48

and Singapore. Instead of competing

16:50

with music streaming services, TikTok will drive

16:53

users to them via its add to

16:55

music app feature, the company said in

16:57

a statement on Tuesday. The feature, which

16:59

was introduced last year, lets users save

17:01

tracks they like on TikTok to playlists

17:03

on partner services such as Spotify, Apple

17:05

Music, and Amazon, end quote. So

17:15

there is a big meta event tomorrow at

17:17

1 Eastern. I'm not

17:19

sure what I'm gonna do about

17:21

that yet. You'll either get the show way early if

17:24

I feel like there's enough news that I

17:26

don't want to get crowded out, but

17:28

I think chances are I will wait to cover

17:31

the event itself and the show will be a

17:33

bit later than usual. We'll see. Talk

17:35

to you then.