### 어린 사용자를 위한 Roblox의 새로운 안전 조치

### Roblox의 안전 강화: 사용자 보호를 위한 새로운 조치
* 👨 "새로운 자녀 보호 기능, 업데이트된 커뮤니케이션 방법, **13세 미만 사용자를 위한 추가 보호 기능** 등 어린 사용자의 안전을 개선하기 위한 '세 가지'를 출시했습니다.
* 🔍 Roblox는 개인정보 보호 문제로 인해 연령 확인에 어려움을 겪고 있지만, 가입 시 명시된 연령에 관계없이 **사용자 안전을 우선시**합니다.
* 🛡 플랫폼은 모든 커뮤니케이션을 엄격하게 모니터링하고 정책 위반을 해결하고 안전을 보장하기 위해 **전담 조사팀**을 운영하고 있습니다.
* 🤖 AI와 머신 러닝의 비약적인 발전으로 특히 음성 커뮤니케이션을 위한 자동화된 중재 시스템이 향상되어 **로블록스는 안전 기술 분야의 리더**로 자리매김했습니다.
* 👮 전 세계적으로 약 9천만 명의 사용자를 보유한 Roblox는 **행동 이상 징후를 적극적으로 감지**하고 업로드되는 모든 콘텐츠를 검토하여 안전한 환경을 유지합니다.
* 🚨 악플러의 비율은 낮지만, 2018년 이후 이 플랫폼에서 악용과 관련하여 **최소 24명이 체포**되어 지속적인 경계가 필요함을 강조하고 있습니다.
* ⚖ Roblox는 성장과 안전의 균형을 유지하며 잠재적인 위험을 해결하기 위해 **새로운 기능 출시 속도를 늦추는 등 안전을 최우선 과제**로 삼고 있습니다.
* 👨 사용자 안전에 대한 책임은 플랫폼과 부모가 공유하며, **Roblox는 부모가 없을 때에도 사용자를 보호하기 위한 사전 조치**를 취합니다.

Roblox는 고급 AI 시스템과 전담 팀을 통해 사용자 안전을 최우선시하며 악의적인 행위자에 효과적으로 대응합니다.

Just go into some of the measures in
particular that you think will make a real difference here.
First, thank you for having me. We've launched three things.
The first is new parental controls that make it easier for parents to monitor
and manage their kids' accounts even when

they're not sitting next to their kids.
We've also updated how our younger users under 13 can communicate on the
platform. And finally, we've built in additional
protections for our youngest users. Let's just talk about ultimately
defining and working out who is 13 and

under.
It's a really difficult list. If people aren't going to be always
telling the truth or using technology here, or is it really, depending on the
truth of the user base? You know, if you're under 13 and you
sign up for an account online, we

actually cannot ask you for additional
information because of privacy reasons. So from a Roblox perspective, we work to
make sure that everybody is on the platform is safe no matter what age
they sign up at. So how therefore, are you looking to
kind of ask if you cannot get the data

and that clarity is there is there a
fix? Is there something you're working on?
Well, so what we do is we take extraordinary steps to keep everybody
safe on the platform. For example, when you're communicating
on Roblox, we don't encrypt anything.

We monitor all communication and we have
an investigation team that's constantly looking into issues where our systems
detect that something may be violative to our process or our policies.
And we have a lot of machine learning and AI systems that are running to keep
everybody safe.

And really, that's part of our heart.
Oh, go ahead, talk through that. AI, the machine learning, how
sophisticated is it becoming? Because the subtleties of grooming are
so hard to pick up on.

That's true.
AI is advancing very fast and we've made significant leaps in the technology.
For example, our automated moderation systems for voice communication.
Just earlier this year, we open sourced some of the models behind that
technology, and there are some of the

most popular models in open source today
for safety. So we feel that Roblox is really at the
leading edge of developing AI to keep people safe on the platform.
Can you give that balance of AI and then the team that does the deeper
investigation?

Is it that you can ever more focus on
technology being the fix here? How big is the team going to remain?
Really, everything is almost like a multilayered approach.
You have AI systems that make initial decisions and act very fast at its
super scale.

But behind those eyes, you have a team
of professional safety experts that are doing all sorts of things.
They're the ones who are, of course, who are training AI.
They're also the ones that are handling appeals because sometimes the AI
doesn't get things right.

So if somebody says, Hey, I feel like it
made a mistake, we have our team of professionals who step in to make
decisions. And of course, that trains the AI.
And then we also have that team of professionals handling the most
complicated cases.

So, for example, we have a team of
investigators who are constantly looking at the platform, trying to understand if
there are bad actors and if there are, immediately removing them from the
platform and working with our law enforcement partners to keep the
platform safe.

They're focused on the bad
actors side. You've outlined how you're going to be
protecting the users, the children, but the people who come on, who aren't who
they say they are, who are indeed the

predators.
What are you doing to innovate there? Sure.
So let me step back for a minute and say, you know, Roblox has nearly 90
million users from 190 countries around the world.
And these users are enjoying experiences

with their friends and with their
families. And it's a really important part of
their life. There's a very, very small number of
people in the world who are just they're bad actors.
And so we have systems that

automatically detect behavioral
anomalies to determine when something might be violating our policy.
And we try and act as quickly as we can. That notion of acting as quickly as we
can actually applies to how people upload content to Roblox.
First of all, when you're building a

game on Roblox, we review every
piece of content that's uploaded to the platform before anybody sees it.
And when you're communicating using text on the platform, we actually block out
and filter a lot of that communication before it ever transmits to other people
if we find that it has violated our

policies.
So those sorts of changes that you just outlined, are they new and novel?
Because I mean, very sadly, as you say, there are very few bad actors in a
percentage perspective, but raw numbers of at least 24 people being arrested
since 2018 because of abusing or

abducting victims through the platform.
Is it 24 too many, as I'm sure you agree?
I think anybody obviously who is comes to any harm on Roblox is one too
many. And we take these situations incredibly
seriously and we do everything we can to

keep the platform safe.
It is our number one priority when it comes to the bad actors on
Roblox. We are constantly innovating and we've
actually this year alone have already had 30 major updates to our safety systems.
So the release that we're talking about

today and that we announced this week is
really about limiting exposure to some of our youngest audiences to content
that's more appropriate to teenagers when we're talking about the actual the
really bad actors. We are constantly innovating on that
front and constantly updating our

systems.
You talk about safety being your number one priority, and there have, of course
been accusations that there was growth that came first over safety.
Do you still feel in any way that that's the case or do you ultimately have to
slow down sign ups in some way to ensure

the number one priority is safety?
You know, our number one priority is safety.
I mean, I think that's just that our first principle and if you think about
it from a company perspective, Roblox was founded nearly 20 years ago where
the audience was primarily kids in the

beginning and really since its
foundation, protecting the users on the platform has been our first priority.
And it's good for our business, it's good for growth, it's obviously good for
our users. It everyone on Roblox understands that,
yes, sometimes we slow things down to

account for safety issues.
For example, when we first start developing new features, we'll sit down
with those feature teams and talk about risks associated with a feature and make
sure we address those at the very beginning rather than waiting until the
very end and trying to fix something.

We we built safety and from comfort at
the beginning, you're not the only company that in many ways has let a
genie out of a bottle here because of those few bad actors.
And I'm interested is ultimately how the how the focus shifts from being one
where companies are the ones we turn to

to fix this, or how much of the onus is
on the parents, on the children, on the users?
Where do you think that balance lies, Matt?
I think companies need to to take a proactive stance to keeping their users
safe on the platform.

Obviously, we want to provide as many
systems and controls and insights to parents as possible.
But from our perspective, we need to make the assumption that parents aren't
always in the room and aren't always involved.
And that's okay.

Parents are really busy.
They're, you know, spread between their jobs and all kinds of different things
going on. And we need to play a proactive role.
I think when you're talking about like the larger problem that we face, I think
it's almost like it's it's a problem

across the internet.
And so the other thing that we think a lot about on Roadblocks is how do we
keep people safe on Roadblocks and knowing that our users also have
accounts on other platforms and they have phone numbers and cell phones and
things like that, and we do our best to

keep people on Roadblocks.
So we try and block people from sharing personally identifiable information.
We try and block people from directing users to other platforms because we know
that we can keep Roadblocks safe and we know that we go to great lengths to keep
the platforms safe.

But other platforms don't operate at the
same standards that we do. So it's very important for us to do what
we can to make sure that people stay safe on Roadblocks while they're there.

Roblox Issues New Safety Features

Welcome to the Bloomberg Technology YouTube channel. Subscribe for original series and mini-docs on game-changing technologies, profiles on innovators, live conversations with experts, and breaking technology news & analysis from Bloomberg TV.

Bloomberg Technology

New parental controls and platform updates for user safety

Introduction of new parental controls for easier monitoring of children's accounts.

Updates on communication methods for users under 13.

Additional protections implemented for the youngest users.

Challenges in verifying user age due to privacy regulations.

Roblox's safety measures: AI, monitoring, and investigation team

Unencrypted communication for monitoring purposes.

Use of machine learning and AI systems for safety.

Advanced AI for detecting subtle grooming behaviors.

Open-sourced AI models for voice communication moderation.

Balance between AI and human investigation teams.

Multilayered approach with AI for initial decisions and human experts for appeals and complex cases.

Human team training the AI and handling appeals.

Team of investigators actively monitoring for and removing bad actors.

Addressing bad actors and platform safety

Vast user base and the relatively small number of bad actors.

Systems for automatically detecting behavioral anomalies.

Content review before publication and communication filtering.

Addressing concerns about the number of arrests related to platform abuse.

Commitment to platform safety and continuous innovation.

Over 30 major updates to safety systems this year.

Focus on limiting exposure of young audiences to inappropriate content.

Addressing accusations that growth prioritized over safety.

Safety as the company's top priority since its founding.

Slowing down processes to address safety concerns.

Addressing safety risks at the beginning of feature development.

Discussion on the balance of responsibility between companies, parents, and users.

Companies' proactive role in user safety.

Providing tools and controls for parents, while acknowledging their limitations.

Addressing the broader internet-wide problem of online safety.

Efforts to prevent sharing personal information and directing users to other platforms.

Maintaining platform safety while acknowledging varying standards across different platforms.