WEBVTT

00:00:00.180 --> 00:00:04.080
So,
Open Claw or Clawed Bot as it used to be

00:00:04.140 --> 00:00:07.750
and Mold book,
it's been some intense days on this

00:00:07.780 --> 00:00:11.720
We got a new AI hype and, of course,
I also spent the

00:00:11.760 --> 00:00:15.640
last days trying to get as much as
possible out of

00:00:15.700 --> 00:00:19.190
Open Claw, and I got some feelings
and thoughts, and they

00:00:19.360 --> 00:00:23.300
differ from most of the other videos
and posts

00:00:23.860 --> 00:00:27.720
I saw.
But let me start with a short story,

00:00:27.780 --> 00:00:31.680
and I'm sure you can figure out the
analogy.

00:00:31.720 --> 00:00:35.460
Imagine you are living in a town,
a village, and in that

00:00:35.560 --> 00:00:39.220
town,
there is that really friendly guy that

00:00:39.500 --> 00:00:42.680
eager and happy to help you with all kinds
of

00:00:42.740 --> 00:00:46.700
tasks.
He does all the chores you don't want to

00:00:46.740 --> 00:00:50.700
at least he tries to.
He's happy to take your kids

00:00:50.710 --> 00:00:54.160
to school, to clean your house,
clean your car, do the

00:00:54.200 --> 00:00:58.080
groceries for you, and you can lean back,
relax,

00:00:58.160 --> 00:01:01.860
and let's say, um,
in order to really help you, of

00:01:01.960 --> 00:01:05.720
course, that assistant,
that person needs broad

00:01:06.000 --> 00:01:09.180
permissions.
You need to give him the keys to your

00:01:09.220 --> 00:01:13.160
You need to give him the keys to your car
so that he can clean

00:01:13.180 --> 00:01:16.300
it from the inside
and do groceries with it.

00:01:16.380 --> 00:01:19.920
You also, of course,
need to tell your kids to get into the

00:01:19.980 --> 00:01:23.870
car with him so that he can take them to
school, and so on

00:01:23.940 --> 00:01:27.520
and so forth. Now, there
is a problem with that guy

00:01:27.560 --> 00:01:31.140
though. He's super friendly,
but he sometimes comes to weird

00:01:31.240 --> 00:01:35.000
conclusions. At least,
you can't rule out that he will

00:01:35.140 --> 00:01:38.960
come to weird conclusions.
He may conclude that

00:01:39.000 --> 00:01:42.980
the best way to get rid of all the dirt in
your house is to set

00:01:43.000 --> 00:01:46.800
it on fire. Unfortunately, he also is

00:01:46.900 --> 00:01:50.720
easily influenced by others, at least
if they're a

00:01:50.760 --> 00:01:54.720
bit more deceptive about it. He can

00:01:54.780 --> 00:01:58.220
be influenced to maybe steal your car

00:01:58.380 --> 00:02:02.200
because that's better for society as a

00:02:02.260 --> 00:02:06.070
whole. Again, not guaranteed, not

00:02:06.120 --> 00:02:09.360
necessarily going to happen,
but absolutely possible.

00:02:09.419 --> 00:02:13.360
You can't rule it out. And, therefore,

00:02:13.400 --> 00:02:17.040
of course,
you unfortunately have to take away

00:02:17.360 --> 00:02:21.180
many of the permissions
and much of the access you granted

00:02:21.280 --> 00:02:24.980
that guy because you can't entirely trust

00:02:25.020 --> 00:02:28.660
him, and the things that could happen
are too

00:02:29.080 --> 00:02:32.540
bad for you to just live with them or

00:02:32.549 --> 00:02:35.130
accept them as a potential danger.

00:02:35.160 --> 00:02:38.989
So unfortunately, of course,
as you take away many of those permissions

00:02:39.080 --> 00:02:43.000
access rights, he gets less
and less useful

00:02:43.160 --> 00:02:46.340
to you. And then there is another problem.

00:02:46.440 --> 00:02:50.000
Even with broad permissions, you didn't

00:02:50.160 --> 00:02:53.820
get as much use out of him as you hoped

00:02:53.840 --> 00:02:57.700
to because the tasks you were promised he

00:02:57.740 --> 00:03:01.540
could do, he only sometimes did,
and some of them he

00:03:01.640 --> 00:03:05.400
was not able to do at all
or he forgot how to do

00:03:05.440 --> 00:03:09.320
something or did the same task differently
every time you

00:03:09.460 --> 00:03:13.040
asked him about it
or needed a lot of input from your

00:03:13.180 --> 00:03:17.060
side. So ultimately, you're not convinced,

00:03:17.100 --> 00:03:20.780
and that's been, you guessed it,
my experience with

00:03:21.080 --> 00:03:24.520
Open Claw, and believe me, I, I tried.

00:03:24.600 --> 00:03:28.590
I read many good things.
I heard many good things about it,

00:03:28.640 --> 00:03:32.260
tried. I spun up my own VPS. By the way,
uh, I didn't

00:03:32.420 --> 00:03:36.200
know, but you can actually also use VPS,
uh,

00:03:36.220 --> 00:03:39.440
providers other than Hostinger.
Nothing against Hostinger.

00:03:39.520 --> 00:03:43.400
Uh, I just had a different feeling
when I watched many of those videos,

00:03:43.740 --> 00:03:47.640
but anyways. Uh, I spun up my own VPS and

00:03:47.800 --> 00:03:51.420
I installed Open Claw on it,
and of course, you could also

00:03:51.460 --> 00:03:55.080
install it on, on your system. Um,
there is one single

00:03:55.140 --> 00:03:58.880
command you need to run and, uh,
then you're good to go, but

00:03:58.960 --> 00:04:02.460
personally, I would never install it on my

00:04:02.480 --> 00:04:06.400
MacBook even though I'm fully aware
that I would be able

00:04:06.440 --> 00:04:10.380
to get more out of it
if I would install it there,

00:04:10.420 --> 00:04:14.010
why I didn't install it there
and why I would never install it there

00:04:14.040 --> 00:04:17.930
now, uh, later.
So I installed it on my VPS and I

00:04:17.940 --> 00:04:21.600
went through that onboarding flow,
and I'm sure you saw that many times now

00:04:21.640 --> 00:04:25.220
already and you maybe already went through
it on your own.

00:04:25.480 --> 00:04:29.360
I linked it up to my ChatGPT Plus
subscription in the

00:04:29.440 --> 00:04:33.300
end. I set up my Telegram bot and I

00:04:33.360 --> 00:04:37.240
was ready to communicate with my bot in
the end, with my

00:04:37.300 --> 00:04:41.240
Open Claw bot. And then there I

00:04:41.300 --> 00:04:44.800
sat and had to think of things

00:04:45.380 --> 00:04:49.310
I wanted it to do for me. Now, of course,
I've seen plenty of

00:04:49.380 --> 00:04:53.000
other posts and videos where people used
it

00:04:53.080 --> 00:04:56.920
to have it build dashboards for them

00:04:57.100 --> 00:05:00.540
or do web research or find cheaper

00:05:00.720 --> 00:05:04.670
flights or even buy stuff, but

00:05:04.900 --> 00:05:08.469
I didn't feel like giving it access to my
credit card, and,

00:05:08.540 --> 00:05:12.300
um, I, I'm not sure about you
but I typically don't fly

00:05:12.340 --> 00:05:15.600
three times a day, so, um,
looking for those flights

00:05:15.680 --> 00:05:19.580
myself, especially since there
are flight comparison sites out there

00:05:19.640 --> 00:05:22.746
that find you the cheapest flight,
wasn't.....

00:05:22.776 --> 00:05:26.696
too difficult of a task for me
and I genuinely enjoy the process

00:05:26.756 --> 00:05:30.076
of planning my trips. But, of course,
that may be different for everybody else.

00:05:30.096 --> 00:05:33.976
Now, for research, I had the problem
that I'm super happy with the

00:05:34.056 --> 00:05:37.796
AI-powered research tools
that already exist, like the AI

00:05:37.896 --> 00:05:41.616
mode, uh, on google.com
or Deep Research on

00:05:41.696 --> 00:05:44.416
Gemini or on ChatGPT. I use those a lot.

00:05:44.456 --> 00:05:48.256
I find them really helpful so I didn't
really need my

00:05:48.316 --> 00:05:51.946
own bot for that that has a high

00:05:52.096 --> 00:05:54.236
chance of performing worse actually.

00:05:54.256 --> 00:05:57.396
Now, I do get there
are certain areas where it

00:05:57.456 --> 00:06:00.416
could be better than those other

00:06:00.496 --> 00:06:03.856
research, uh, bots or services. For

00:06:03.956 --> 00:06:07.736
example,
if I would grant it access to my X

00:06:07.816 --> 00:06:11.556
say, um, I understand that it could,
of course, do

00:06:11.616 --> 00:06:15.436
research in areas where you need to be
logged in or

00:06:15.556 --> 00:06:19.416
where my history matters. I,
I fully get that,

00:06:19.496 --> 00:06:23.456
um,
so that's why I'm using SuperGrok for

00:06:23.556 --> 00:06:27.336
if I wanna research on X. But yeah,
I get that if you

00:06:27.396 --> 00:06:30.916
give it broad permissions,
if you allow it to log into your

00:06:30.976 --> 00:06:34.396
accounts, use your browser,
maybe run on your system, you

00:06:34.476 --> 00:06:37.976
can probably get a bit more out of it than
I was

00:06:38.096 --> 00:06:41.496
able to get out of it.
And maybe I'm just also not

00:06:41.776 --> 00:06:45.696
creative enough. And by the way,
just to be very clear, and I

00:06:45.716 --> 00:06:49.556
think I have made
that clear in other videos too,

00:06:49.656 --> 00:06:52.636
of AI, not just for research
but also for coding.

00:06:52.656 --> 00:06:56.556
For example,
I recently released an entire Claude Code

00:06:56.596 --> 00:07:00.556
I'm using Claude Code
and all these other tools like Cursor for

00:07:00.836 --> 00:07:04.656
software. I think AI is a huge help

00:07:04.736 --> 00:07:08.556
there or can be a huge help there.
So that's not a general

00:07:08.596 --> 00:07:12.056
thing against AI.
I just genuinely didn't find

00:07:12.066 --> 00:07:15.996
the amazing use cases for OpenClaire,
especially when not

00:07:16.056 --> 00:07:19.606
running it on my machine, and that is

00:07:19.656 --> 00:07:23.206
the main problem I actually have with it.

00:07:23.216 --> 00:07:27.026
Because you could definitely say
that I'm just not creative enough or not

00:07:27.076 --> 00:07:30.666
open-minded enough to find the right use
cases for it,

00:07:31.456 --> 00:07:35.316
but security is a huge issue

00:07:35.496 --> 00:07:39.356
I have with OpenClaire.
And I know there are people

00:07:39.396 --> 00:07:43.356
that will tell you
that they used it for weeks

00:07:43.376 --> 00:07:47.356
wrong or that this, uh, will all,
of course, get

00:07:47.396 --> 00:07:50.856
better,
and I will say the first argument

00:07:51.096 --> 00:07:55.056
wrong, um, well,
that's not the kind of argument

00:07:55.096 --> 00:07:58.756
that convinces me because just because
nothing went wrong

00:07:59.096 --> 00:08:02.896
for you does not mean that nothing is

00:08:02.936 --> 00:08:06.336
going wrong in general and
that there wouldn't

00:08:06.396 --> 00:08:10.286
be huge security issues that

00:08:10.396 --> 00:08:14.336
can, of course,
be exploited by bad actors or

00:08:14.376 --> 00:08:17.976
that, of course,
things could simply go wrong because

00:08:18.096 --> 00:08:22.046
AI, large language models,
is unpredictable.

00:08:22.076 --> 00:08:25.806
Of course,
the chance for it erasing your hard drive

00:08:25.836 --> 00:08:29.616
extremely high. It's super low
but it's not zero

00:08:29.676 --> 00:08:33.656
and it will never be zero with large
language models

00:08:33.696 --> 00:08:37.546
without additional checks.
They can be unpredictable.

00:08:37.635 --> 00:08:41.286
In addition,
in the official security documentation of

00:08:41.336 --> 00:08:45.056
OpenClaire,
they are correctly stating that prompt

00:08:45.176 --> 00:08:48.976
injection is not solved. Of course, the

00:08:49.076 --> 00:08:52.596
latest models like GPT-5.2
and so on got much

00:08:52.776 --> 00:08:56.576
better at protecting against prompt
injection.

00:08:56.596 --> 00:09:00.396
They got much better at following
instructions,

00:09:00.496 --> 00:09:04.066
on. But there is no 100% protection

00:09:04.096 --> 00:09:08.056
against, uh, prompt injection
and the way large language models

00:09:08.156 --> 00:09:11.516
work, there never will be.
So prompt injection

00:09:11.576 --> 00:09:15.056
attacks can't be ruled out and, of course,
the

00:09:15.536 --> 00:09:18.696
more popular tools like OpenClaire get,
the more

00:09:18.776 --> 00:09:22.596
people that are running it,
the more it will be in the

00:09:22.696 --> 00:09:26.236
focus of bad actors. And there are various

00:09:26.316 --> 00:09:30.196
ways of injecting prompts into an active

00:09:30.376 --> 00:09:34.116
OpenClaire bot because you may think,
"Well, I'm the only one

00:09:34.136 --> 00:09:37.896
communicating with it.
I have my Telegram bot set up and

00:09:37.976 --> 00:09:41.336
only I have access to
that so I'm safe." Well, think

00:09:41.456 --> 00:09:45.415
again. For example, there
is this idea of skills

00:09:45.456 --> 00:09:49.206
with OpenClaire and you may already know
skills from coding agents

00:09:49.236 --> 00:09:52.045
like Claude Code. The idea
is kind of the same.

00:09:52.076 --> 00:09:55.576
The idea is that you expose extra

00:09:55.676 --> 00:09:59.236
context in the end,
an extra Markdown document, though

00:09:59.276 --> 00:10:02.976
potentially also coupled with executable
scripts, uh,

00:10:02.986 --> 00:10:05.736
to the agent to give it more capabilities.

00:10:05.776 --> 00:10:09.766
So for example, to, uh,
give it some extra documentation on how

00:10:09.856 --> 00:10:13.246
to interact with Slack here in this
example.

00:10:13.246 --> 00:10:16.956
And then as mentioned,
a skill can also come bundled up with some

00:10:17.016 --> 00:10:20.616
additional script
which the AI agent can execute to

00:10:20.656 --> 00:10:24.636
efficiently do something like generate an
image or send a message to Slack

00:10:24.676 --> 00:10:28.306
or whatever it is Now the problem is that

00:10:28.496 --> 00:10:31.616
Claw Hub, the official skills hub for

00:10:31.656 --> 00:10:35.415
OpenClaire,
initially at least allowed everybody

00:10:35.476 --> 00:10:38.906
to submit skills. So it was pretty

00:10:38.956 --> 00:10:42.415
easy to run supply chain

00:10:42.436 --> 00:10:45.256
attacks which we saw from the npm

00:10:45.396 --> 00:10:48.996
ecosystem, uh, last year, uh,
totally unrelated to

00:10:49.076 --> 00:10:52.856
AI, which essentially means that, uh,
a bad

00:10:52.936 --> 00:10:56.915
actor can publish a skill
that tells the AI

00:10:56.976 --> 00:10:59.915
to do something bad and that
is just a prompt injection.

00:10:59.956 --> 00:11:03.836
So just by installing a malicious skill,
you could

00:11:03.856 --> 00:11:07.336
expose your agent to a prompt injection
attack.

00:11:07.696 --> 00:11:11.616
Now some fixes were implemented here so,
uh, at

00:11:11.636 --> 00:11:15.596
the point of time where I'm recording
this,

00:11:15.696 --> 00:11:18.556
so the security was vastly improved here.

00:11:18.576 --> 00:11:22.436
But if we learned anything from the supply
chain attacks on npm

00:11:22.506 --> 00:11:26.286
last year, it is
that we definitely can't rule out

00:11:26.415 --> 00:11:29.896
that this skills feature,
this hub can be used to

00:11:29.936 --> 00:11:32.930
inject, um, malicious instructions into...

00:11:33.080 --> 00:11:36.630
into the ecosystem and into your, uh,
OpenClaw

00:11:36.880 --> 00:11:40.300
setup potentially.
And that's not the only way of running

00:11:40.340 --> 00:11:44.270
attacks.
If your bot reaches out to the internet,

00:11:44.320 --> 00:11:47.980
it most likely does, it will, of course,
visit websites or

00:11:48.120 --> 00:11:51.820
read content from websites. And there,

00:11:51.900 --> 00:11:55.720
we also can have malicious websites
that trick the

00:11:55.780 --> 00:11:59.630
AI into following instructions, prompts,
that are

00:11:59.660 --> 00:12:03.360
embedded on that website.
Every piece of text

00:12:03.700 --> 00:12:07.360
your bot reads and processes is a

00:12:07.640 --> 00:12:11.360
prompt in the end,
so every website it visits

00:12:11.760 --> 00:12:15.740
is a prompt, uh, or contains a prompt
that it

00:12:15.820 --> 00:12:19.700
can follow and execute.
And then we got other potential sources as

00:12:19.760 --> 00:12:23.429
well like, for example, emails.
If you use your, uh,

00:12:23.600 --> 00:12:27.320
bot to process incoming emails,
every email, of

00:12:27.400 --> 00:12:30.680
course, acts as a prompt. So prompt

00:12:30.820 --> 00:12:34.560
injection is a, a serious, huge risk
and just

00:12:34.580 --> 00:12:38.120
because nothing went wrong for you ever,
doesn't mean

00:12:38.460 --> 00:12:42.280
things can't go wrong. Now you may,
of course, say, "Well,

00:12:42.420 --> 00:12:46.340
I'm running my bot on a VPS." Or maybe

00:12:46.400 --> 00:12:50.060
you're using something like MaltWorker,
which is in the end a

00:12:50.120 --> 00:12:53.770
pre-built blueprint or setup provided by

00:12:53.880 --> 00:12:57.680
Cloudflare,
which uses various Cloudflare services for

00:12:57.800 --> 00:13:01.660
hosting and running, uh, OpenClaw,
and you should be doing

00:13:01.720 --> 00:13:05.110
that. You should be doing that. Uh,
you should absolutely

00:13:05.320 --> 00:13:09.260
not run it on, on your system.
And there also are

00:13:09.460 --> 00:13:13.300
features like sandboxing, so that
is actually

00:13:13.320 --> 00:13:16.620
built into OpenClaw. They have a-
an entire

00:13:16.630 --> 00:13:20.580
documentation article about sandboxing
and how you can make sure

00:13:20.620 --> 00:13:24.540
your agents run in a sandbox,
which essentially is a darker container,

00:13:24.620 --> 00:13:27.080
so that the blast radius is reduced.

00:13:27.220 --> 00:13:31.140
By the way, the documentation, it's a lot,
but it's

00:13:31.280 --> 00:13:35.250
not good. I spent hours, literally many,
many

00:13:35.400 --> 00:13:39.140
hours trying to secure my setup.
And I'm sure it's all in

00:13:39.180 --> 00:13:42.319
there, and I saw the security article.

00:13:42.360 --> 00:13:45.960
It's just so, so hard.
And before you tell me that I should've

00:13:46.080 --> 00:13:49.720
asked my OpenClaw bot, I did a lot.
It sometimes

00:13:49.780 --> 00:13:52.900
worked, it sometimes didn't. It
was a lot of trial and error.

00:13:52.960 --> 00:13:56.600
So yeah, the documentation and how

00:13:56.700 --> 00:14:00.500
hard it is to get useful information out
of it is

00:14:00.580 --> 00:14:03.880
its own problem, but, of course, one
that can be fixed.

00:14:03.889 --> 00:14:07.760
And I appreciate the fact
that at least the information

00:14:07.800 --> 00:14:11.660
here, just to be clear. But yeah,
so sandboxing is

00:14:11.800 --> 00:14:15.760
built in and is available
and allows you to

00:14:16.160 --> 00:14:19.580
reduce the blast radius, which is super

00:14:19.620 --> 00:14:23.500
important. Uh, because in the end,

00:14:23.580 --> 00:14:26.949
due to the prompt injection, uh,

00:14:26.980 --> 00:14:30.400
vulnerabilities that exist
that can't really be

00:14:30.440 --> 00:14:33.850
solved, reducing the blast radius is

00:14:33.880 --> 00:14:37.680
important. So, for example,
if you use sandboxing,

00:14:37.820 --> 00:14:41.700
overall setup on a VPS,
the worst thing that could happen

00:14:41.800 --> 00:14:45.680
is that, of course,
the stuff in the sandbox gets deleted

00:14:45.760 --> 00:14:49.630
or, depending on your setup,
maybe your entire VPS but not

00:14:49.660 --> 00:14:53.640
your system.
That's the reason why I would never run,

00:14:53.720 --> 00:14:56.480
OpenClaw on, on my machine,
on my main machine.

00:14:56.620 --> 00:15:00.480
I'm-
I absolutely don't want it to erase files,

00:15:00.500 --> 00:15:04.150
whatever, on my machine. So yeah,
reducing the blast radius is

00:15:04.180 --> 00:15:08.120
important. Unfortunately, though,
that still doesn't protect you against the

00:15:08.160 --> 00:15:11.760
worst things that could happen because
with prompt injection

00:15:11.800 --> 00:15:15.540
attacks, of course,
an attacker could try to delete files on

00:15:15.580 --> 00:15:19.560
system. But even worse than that,
they could steal stuff.

00:15:19.620 --> 00:15:23.140
So data exfiltration

00:15:23.260 --> 00:15:26.240
is, in my opinion, a, a bigger problem

00:15:26.680 --> 00:15:30.000
than an attacker deleting files on your

00:15:30.060 --> 00:15:33.400
system. And data exfiltration is

00:15:33.540 --> 00:15:37.320
100% something that can happen or

00:15:37.360 --> 00:15:41.120
that can be the result of a prompt
injection attack because, of course, an

00:15:41.180 --> 00:15:45.030
attacker could get the AI to gather all
the secrets it knows,

00:15:45.120 --> 00:15:49.100
all the passwords it knows,
and it needs to know some passwords

00:15:49.120 --> 00:15:51.070
in order to use your email account.

00:15:51.100 --> 00:15:54.730
Maybe it's-
maybe you gave it your credit card number,

00:15:54.859 --> 00:15:58.220
access to various pieces of data and
that data could be

00:15:58.260 --> 00:16:02.090
collected due to a prompt injection attack
and could

00:16:02.120 --> 00:16:05.710
be exfiltrated, and that is a, uh,

00:16:05.920 --> 00:16:09.890
bigger, uh,
risk than it potentially deleting your

00:16:09.920 --> 00:16:12.330
hard drive if you set it up correctly.

00:16:12.340 --> 00:16:14.220
Of course, it could also do other things.

00:16:14.240 --> 00:16:18.200
It could turn your VPS into a, a bot

00:16:18.520 --> 00:16:22.440
for DDoS attacks, for example,
so that's just

00:16:22.500 --> 00:16:26.220
one example. There
is an endless amount of things

00:16:26.260 --> 00:16:30.070
course, but the main thing to take away
is that through

00:16:30.100 --> 00:16:33.900
prompt injection attacks,
attackers could take over

00:16:34.080 --> 00:16:36.340
your bot and, therefore, your machine.

00:16:36.360 --> 00:16:39.490
They could get your bot to install
malicious software

00:16:40.060 --> 00:16:44.020
to tweak the system configuration
depending on the access rights it

00:16:44.040 --> 00:16:47.700
has, of course,
and then they could potentially take over

00:16:47.800 --> 00:16:51.260
your VPS, your machine. These
are the kind of things that could

00:16:51.340 --> 00:16:54.320
happen. So access rights are the

00:16:54.360 --> 00:16:58.320
important, uh, keyword here,
and sandboxing is

00:16:58.400 --> 00:17:01.680
one crucial part in that.
It's not all though.

00:17:01.760 --> 00:17:05.699
You can configure...... your OpenClaw

00:17:05.780 --> 00:17:09.520
bot such that it has to ask for approval
when

00:17:09.599 --> 00:17:13.159
running in sandbox mode, at least,
for executing

00:17:13.280 --> 00:17:17.099
certain tasks. But
that kind of defeats the idea of

00:17:17.260 --> 00:17:21.099
having a bot that runs behind the scenes
and does stuff whilst you are

00:17:21.109 --> 00:17:24.579
away because you all the time have to give
it

00:17:24.599 --> 00:17:27.939
approval for all the kind of stuff it
wants to do

00:17:27.980 --> 00:17:30.860
suddenly, and that, of course,
gets super annoying.

00:17:30.919 --> 00:17:34.899
So you just might not read anymore what
it's asking approval for, you

00:17:34.939 --> 00:17:38.320
might always grant approval,
and at some point,

00:17:38.379 --> 00:17:41.879
just annoys you. Because again,
it's not really useful if you have to

00:17:41.959 --> 00:17:45.379
manually approve everything.
So combine that,

00:17:45.419 --> 00:17:49.219
combine these security issues and the fact

00:17:49.280 --> 00:17:52.800
that I did not find a way of running this

00:17:52.860 --> 00:17:56.419
securely in a way I would feel good with,
with the

00:17:56.500 --> 00:18:00.199
fact that I didn't really find those super
amazing

00:18:00.300 --> 00:18:03.800
use cases, combine these things
and you end up with a

00:18:03.840 --> 00:18:07.139
situation where, uh,
I'm just not using OpenClaw

00:18:07.219 --> 00:18:10.949
anymore. And of course,
that can be different for you

00:18:10.980 --> 00:18:14.560
people that were super excited, and yeah,
it's possible that the

00:18:14.600 --> 00:18:18.310
future of personal AI-powered assistants
looks

00:18:18.310 --> 00:18:22.240
something like this. It's possible
that better security

00:18:22.280 --> 00:18:25.800
mechanisms can be introduced and can be

00:18:25.939 --> 00:18:29.560
invented that don't require your constant
approval for

00:18:29.780 --> 00:18:33.120
everything or that make
that approval process easier

00:18:33.719 --> 00:18:37.600
and therefore allow you to securely run
assistants like

00:18:37.659 --> 00:18:41.560
this. That is all possible.
I wouldn't rule out that

00:18:41.580 --> 00:18:45.139
this happens, and of course, it
is an impressive

00:18:45.379 --> 00:18:48.780
feat that a single developer built this
tool, though, of

00:18:48.860 --> 00:18:52.439
course,
not looking at the code at all does have

00:18:52.500 --> 00:18:56.179
price, uh, as many bugs and

00:18:56.280 --> 00:18:59.679
security problems, uh,
certainly also show.

00:18:59.740 --> 00:19:03.379
Not that software wouldn't have any
security problems if

00:19:03.980 --> 00:19:07.659
you would review everything,
but it certainly, in my opinion,

00:19:07.699 --> 00:19:11.669
don't look at the code at all.
But nonetheless,

00:19:11.699 --> 00:19:15.689
if you ask yourself the question,
why OpenAI or Google didn't

00:19:15.739 --> 00:19:19.389
come up with a product like this,
the reason may be a lack of

00:19:19.419 --> 00:19:23.179
innovation, but of course,
it's also the fact that a tool like

00:19:23.219 --> 00:19:26.639
this can right now only exist as open
source

00:19:26.699 --> 00:19:29.239
software without any legal

00:19:29.320 --> 00:19:32.980
obligations because this thing is not

00:19:33.120 --> 00:19:37.110
something Google could sell
or run for you with

00:19:37.120 --> 00:19:40.770
broad permissions. But of course,
it's definitely possible that this is

00:19:40.800 --> 00:19:43.620
the initial spark that gives us

00:19:44.000 --> 00:19:47.379
safer, maybe more useful personal

00:19:47.419 --> 00:19:50.120
AI-powered assistants in the future.

00:19:50.399 --> 00:19:54.389
And, uh,
just to also briefly mention Maltbook, uh,

00:19:54.419 --> 00:19:57.659
that is a thing I totally did not
understand.

00:19:57.760 --> 00:20:01.520
Uh, it was meant to be a social network,
a Reddit

00:20:01.580 --> 00:20:05.439
for AIs only. It turned out that it was

00:20:05.580 --> 00:20:09.540
actually very human-orchestrated and, uh,
quite

00:20:09.580 --> 00:20:13.560
a bit fake as I understand it, and it had

00:20:13.639 --> 00:20:17.620
gapping security issues and,

00:20:17.739 --> 00:20:21.600
yeah, I don't know. AI has positive

00:20:21.899 --> 00:20:25.719
use cases or positive implications,
I guess.

00:20:25.739 --> 00:20:28.699
AI has a lot of negative implications.

00:20:28.800 --> 00:20:32.679
Um, this thing here
is not something the world needs

00:20:32.760 --> 00:20:36.460
in my opinion. But yeah, OpenClaw,
definitely

00:20:36.469 --> 00:20:40.208
interesting, maybe super useful for you,
uh,

00:20:40.239 --> 00:20:44.060
definitely not my cup of tea/coffee,

00:20:44.159 --> 00:20:45.939
uh, right now