00:00:00
the government appears to be developing
00:00:01
software using AI to spy on you to
00:00:05
discover what is being referred to as
00:00:07
concerning posts all the while it's also
00:00:11
reported that they might be trying to
00:00:13
rename these or Rebrand these non crime
00:00:16
hate incidents which I've spoken about
00:00:18
in previous videos which I think is
00:00:20
frankly a waste of time because as
00:00:22
others have already commented it's
00:00:23
pretty much the same thing in orbit name
00:00:27
so I move to this arle from the Free
00:00:31
Speech Union if you're not a member I
00:00:33
suggest you go and join if you value
00:00:35
your free speech not a sponsored video I
00:00:37
hasten to add uh they've not asked me to
00:00:39
do this or anything else but here we are
00:00:42
so a very interesting article here about
00:00:44
non-crime hate incidents um potentially
00:00:46
being renamed rather than being
00:00:48
abolished now frankly I think they
00:00:50
should be abolished because if
00:00:51
something's not a crime it's nothing to
00:00:52
do with the police not really anyway
00:00:54
apart from keeping the peace which they
00:00:56
could do in various ways but without
00:00:58
recording anything and going to the
00:01:00
extent of something that might then
00:01:01
later appear on an enhanced DPS check
00:01:03
which could cause you all sorts of
00:01:04
problems even though you've committed no
00:01:05
crime I've spoken about that in previous
00:01:07
videos College of policing however
00:01:09
disagrees they they seem to think that
00:01:12
um a proactive approach would prevent
00:01:15
incidents from escalating into more
00:01:16
serious offenses which on on The Grand
00:01:19
in the grand scheme of things that's a
00:01:20
possibility they could do that but with
00:01:24
so many offenses going uninvestigated
00:01:26
and they just simply don't have the
00:01:28
resources to do that I respectfully
00:01:29
suggest that that's where the time is
00:01:31
better spent now um in theory the police
00:01:34
are only supposed to record these things
00:01:36
where there is a serious risk of
00:01:38
significant harm uh that could then
00:01:41
escalate into criminality not simply
00:01:43
because someone feels offended although
00:01:45
that's how the guidelines are sort of
00:01:46
drafted when somebody perceives
00:01:48
something to be said or done in a sort
00:01:52
of way of hate and someone may be
00:01:54
offended by that then they should be
00:01:56
investigating that and potentially
00:01:58
recording that as a non-crime hate
00:02:00
incident um over 13,000 of these
00:02:03
non-crime hate incidents were recorded
00:02:05
across England and Wales last year
00:02:07
prompting warnings from campaigners and
00:02:09
commentators including myself about the
00:02:12
effect on free speech and I think this
00:02:14
has a very disturbing effect on Free
00:02:16
Speech if we look at a few examples just
00:02:17
recently you'll have seen some from my
00:02:19
recent videos Allison Pearson you'll be
00:02:21
familiar with from the telegraph visited
00:02:24
in at home by Essex police just
00:02:26
following a complaint again over a
00:02:27
social media post and being investigated
00:02:31
well initially believed that she was
00:02:33
being investigated over a non-crime hate
00:02:34
incident uh and this obviously resulted
00:02:37
in much public debate you remember the
00:02:38
grandmother that hit social media with
00:02:40
the video of the police officers in
00:02:42
plain clothes attending her home after
00:02:44
she posted a critical comment over her
00:02:47
local counselor on Facebook saying that
00:02:49
he should be resigning um government has
00:02:52
defended the use of these in saying that
00:02:54
they are combating hate and everything
00:02:56
else which I support I support combating
00:02:59
hate and
00:03:00
stamping that out but is this really the
00:03:02
right way of going about it because as
00:03:04
we know with many other things it's open
00:03:05
to abuse now again this article here
00:03:07
says critics claim that they are a
00:03:09
distraction from overstretched police
00:03:10
forces that's exactly something that
00:03:12
I've been saying for some time now I'm
00:03:14
I'm Not For a Moment suggesting that
00:03:15
anyone is pulling bits from what I'm
00:03:17
saying maybe they are I don't know I
00:03:19
don't know if they watch my videos or
00:03:20
not but I've been saying that I think
00:03:22
that that is a huge distraction from
00:03:23
overstretch police resources and I think
00:03:25
with respect police officers should be
00:03:27
policing real crime and actually going
00:03:29
to sort things out because there are so
00:03:30
many victims of real crimes that don't
00:03:33
get uh police attention that they really
00:03:35
should but then we come to what is this
00:03:39
government doing with AI and I say this
00:03:43
government um it's either this
00:03:44
government the last government a
00:03:46
combination or an ongoing thing because
00:03:47
this appears to have been set up in 2019
00:03:50
obviously some some years ago so um this
00:03:55
again from Free Speech Union here says
00:03:57
that uh it's a controversial this
00:04:00
information team developing a secretive
00:04:03
AI program which is trolls through
00:04:05
social media looking for concerning
00:04:07
posts that it deems problematic that it
00:04:10
might take action on so here's an
00:04:12
extract and I'll link this below so you
00:04:13
can go and read it uh it says record
00:04:15
show the department for science
00:04:16
innovation technology recently awarded a
00:04:18
2.3 million pound contract to faculty AI
00:04:21
to build monitoring software which can
00:04:24
search for foreign intelligence fair
00:04:26
enough um detect deep fakes fair enough
00:04:29
and analyze social media narratives and
00:04:31
that's where it might start to cross a
00:04:33
little bit of a grave threshold um the
00:04:35
platform is part of the counter
00:04:36
disinformation unit which was set up in
00:04:38
2019 hence the 2019 bit um and sparked
00:04:42
widespread criticism for amassing files
00:04:44
on journalists academics and MPS Who
00:04:46
challenged the government's narrative
00:04:47
during the pandemic now lots of people
00:04:50
challenged the government's narrative
00:04:51
during the pandemic you may have been
00:04:53
familiar with that you may have done
00:04:54
that yourself on social media maybe
00:04:56
there's a file on you um even if you're
00:04:58
not a journalist but even if you you may
00:05:00
be a citizen journalist or you may just
00:05:02
be someone sitting at home posting what
00:05:04
you think which as we know can have
00:05:06
consequences maybe there's a file on you
00:05:08
this is perhaps something you might need
00:05:10
to be aware of the unit which has been
00:05:12
rebranded the national security online
00:05:14
information team has links to
00:05:16
intelligence agencies which has allowed
00:05:18
it to avoid public
00:05:19
scrutiny it reports here um the new AI
00:05:23
tool called the counter disinformation
00:05:24
data platform is looking solely for
00:05:26
posts which pose a threat to National
00:05:28
Security and public safety risk
00:05:30
now I would think most things that you
00:05:31
are likely to post are not going to be a
00:05:33
threat to National Security and public
00:05:35
safety risk and any of those things but
00:05:38
how do we know that this AI tool is not
00:05:40
going to pull lots of people who are
00:05:42
completely innocent uh start pulling
00:05:44
files on them put them into a database
00:05:46
get them investigated and suddenly we're
00:05:47
looking at a bit of a a breach of their
00:05:50
right to privacy here potentially the
00:05:52
current focus of beta testing is the
00:05:54
influence of foreign States during
00:05:56
elections which is again is fair enough
00:05:57
and how much do we really want to um
00:06:00
allow this to spill over the edges of
00:06:02
what we think is reasonable I think if
00:06:04
you were to ask the question do we think
00:06:05
it's reasonable that we ensure that
00:06:07
foreign states are not interfering with
00:06:08
our elections we' probably say that's
00:06:09
perfectly reasonable if we were then to
00:06:11
say well is it going to Snoop on social
00:06:13
media look for narratives and highlight
00:06:15
people that it might label a certain way
00:06:17
and say you're saying things wrong
00:06:19
perhaps you might not be so comfortable
00:06:21
with that heavily redacted documents
00:06:24
obtained by Big Brother watch through
00:06:26
Freedom of Information requests show
00:06:27
that the government is reserving the
00:06:28
right to use the platform for other uses
00:06:31
but of course they are um an executive
00:06:33
summary for the project says uh while
00:06:35
the cddp has a current national security
00:06:37
Focus the to has the ability to be
00:06:39
pivoted to focus on any priority area
00:06:42
now there are numerous priority areas
00:06:44
that the government's been looking at
00:06:45
just recently that it might well tune
00:06:47
and focus this new AI power on and you
00:06:51
might be it since 20121 contract show
00:06:54
the government spent more than 5.3
00:06:57
million on developing and uh other
00:07:00
disinformation projects including
00:07:01
detecting coronavirus information
00:07:03
analyzing climate related miss or
00:07:04
disinformation on social media but
00:07:07
there's a problem with that because if
00:07:09
something starts out as misinformation
00:07:11
or disinformation which are subtly
00:07:13
different by the way Mis just meaning
00:07:14
wrong dis meaning to disrupt and uh to
00:07:17
misin to uh change the narrative on
00:07:20
something um if something started out as
00:07:23
being wrong but then actually later on
00:07:26
they make realizations and discoveries
00:07:28
that they actually that was right in the
00:07:31
first place but you've already been
00:07:32
labeled by that point you're already in
00:07:34
a database there's already a file you've
00:07:35
already been investigated and perhaps
00:07:37
you've already been arrested and
00:07:38
questioned maybe even charged and maybe
00:07:40
even imprison potentially at least
00:07:43
there's the potential for this to go
00:07:44
badly wrong later on particularly when
00:07:47
AI is involved um we've had all sorts of
00:07:50
uh interesting things just recently to
00:07:52
do with AI um was an interesting thread
00:07:56
uh just recently I will try to remember
00:07:58
to link to you
00:07:59
um from uh Max scoffield um barister
00:08:03
specializing in uh vat and tax
00:08:06
litigation uh he had an interesting
00:08:08
thread on AI and he suggested and as I
00:08:13
would suggest that no lawyer actually
00:08:15
use this as your primary tool perhaps as
00:08:18
a bit of a guide for some things here
00:08:20
and there but certainly not as the main
00:08:22
guide because it will make things up it
00:08:24
will pull incorrect cases it will I mean
00:08:27
the hallucinations is is a well-known
00:08:29
phenomena for AI making things up making
00:08:31
up cases where it doesn't find one
00:08:33
that's suitable so it makes one up that
00:08:34
makes makes it look like it's suitable
00:08:36
I've read stories of lawyers around the
00:08:38
world have actually submitted documents
00:08:39
and then been disciplined because
00:08:41
they've submitted a document which has
00:08:43
false cases in it so they are citing
00:08:45
authorities that simply don't exist I've
00:08:48
had lots of different experiences with
00:08:51
AI fairly recently uh we've had um and I
00:08:55
don't mind to say because it's not
00:08:57
revealing any details or anything else
00:08:58
but we've had had clients arguing about
00:09:01
something uh using AI going back and
00:09:03
forth and the AI has generated all sorts
00:09:06
of nonsense and misleading and going off
00:09:10
on tangents on different things and
00:09:13
therefore it's been a completely
00:09:15
pointless conversation because it will
00:09:17
pull up almost imaginary arguments and
00:09:21
use a a point of law that's completely
00:09:24
irrelevant to the conversation for
00:09:25
example you might be talking about a
00:09:27
straightforward breach of contract
00:09:28
matter the AI might then suddenly pull
00:09:30
up something to do with negligence and
00:09:32
then start arguing that this is now all
00:09:34
negligent which obviously it isn't
00:09:36
because it was a completely uh
00:09:38
contractual matter in the first place
00:09:40
I've seen LinkedIn articles which I know
00:09:43
are completely written by Ai and there's
00:09:47
some tools like um GPT Z I think it's
00:09:50
called GPT Z again not sponsored but GPT
00:09:53
z uh you can go on that website and uh
00:09:56
paste in articles that you might have
00:09:59
read and it will scan it and give you a
00:10:01
best guess as to whether or not this is
00:10:03
likely generated by Ai and many of these
00:10:06
articles that I've seen on LinkedIn they
00:10:07
look very AI to me so I've dumped some
00:10:09
of them in and sure enough it said 100%
00:10:13
written by Ai and then you get other
00:10:15
machines which are making it more human
00:10:19
like text so it's generated by AI then
00:10:23
it goes through another filter in
00:10:24
another program to make it look like
00:10:25
it's generated by a human so it throws
00:10:27
in natural errors in there to make it
00:10:29
look real um I teach law sometimes at uh
00:10:33
the law school bar students and I've had
00:10:35
students email me asking for a reference
00:10:39
clearly using Ai and I've written back
00:10:42
to say sure I can provide you a
00:10:44
reference but I have to ask is this chat
00:10:46
GPT and after a bit of an awkward uh
00:10:50
break between emails they did reply and
00:10:52
say yes it was chat GPT I just thought
00:10:55
it was better to have a more
00:10:56
professional request for this uh
00:10:58
reference to which obviously I said it's
00:11:01
not a good start because we can spot
00:11:03
these things miles away and I think most
00:11:05
people can spot these AI things miles
00:11:08
away so I I highlight all of that
00:11:11
because there are serious problems with
00:11:13
AI and so there is a huge potential here
00:11:15
for this to go very very badly wrong so
00:11:18
coming full circle if they're using this
00:11:20
software which is utilizing AI to scour
00:11:23
the internet and find uh posts that it's
00:11:27
it thinks is of real concern
00:11:30
and then flag people up for
00:11:31
investigation and put them on a list and
00:11:33
get files pulled and everything else you
00:11:35
may find yourself being questioned by
00:11:37
the police over what may only have been
00:11:39
a parody account or a sarcastic response
00:11:42
to Something in current affairs speaking
00:11:44
of current affairs um a video upcoming
00:11:46
GB news has had a success um in in court
00:11:50
uh with an ofcom decision um broadly the
00:11:54
distinction was between a current
00:11:56
affairs program and a news program more
00:11:58
about that in another another video I'm
00:11:59
still reading the Judgment very very
00:12:01
interesting and I believe my learning
00:12:03
friend Alan Robert shw is doing a video
00:12:04
on that as well I'll link him below he
00:12:06
on Art of law go subscribe to him if you
00:12:09
like his sort of content as well and I'm
00:12:12
sure you'll find that interesting but I
00:12:13
find this uh quite worrying actually
00:12:17
that not only do we have software which
00:12:20
is scouring the internet looking for
00:12:21
things that it might think are
00:12:23
concerning that you've posted and as we
00:12:24
know people have gone to prison for
00:12:26
things that they've posted but also the
00:12:28
obfuscation and the twisting and the
00:12:30
changing of non-crime hate incidents a
00:12:33
Rebrand or something like that quite
00:12:35
simply they shouldn't really exist if
00:12:38
anything the police should you know make
00:12:41
a doorstep make a phone call or try to
00:12:44
keep the peace in some other way not by
00:12:46
recording people in a little black book
00:12:48
which are effectively blacklisted for
00:12:50
certain things it could cause people
00:12:52
serious problems such as if you work
00:12:54
with children you work in a in a school
00:12:55
you work in healthcare or any other sort
00:12:57
of profession if they an enhanced DBS
00:13:00
check and then they see that and they
00:13:01
say ah you've got a hate incident
00:13:04
recorded against you you might lose a
00:13:06
job for that or you might not get hired
00:13:08
for the job for that or you might get
00:13:10
all sorts of other problems so I can see
00:13:11
this as being a real problem especially
00:13:13
if they now combine AI trolling through
00:13:16
posts if they get onto one and they go
00:13:18
down a rabbit hole of your posts and
00:13:19
things that you've done or said online
00:13:21
before even many years ago it might put
00:13:24
you in a file and on a list and I think
00:13:27
that's a problem and I think that's
00:13:28
concerning so let me know what you think
00:13:30
about that in the comments below um I'll
00:13:32
probably do a post about this as well
00:13:33
with a a poll so if I do check that out
00:13:37
in the description and uh with that said
00:13:40
thank you for watching I appreciate you
00:13:42
watching joining me here uh we are on
00:13:44
this journey of understanding law
00:13:46
together and I think it really should be
00:13:47
simplified I think the Law's got far too
00:13:49
complex I think the system is too
00:13:51
complex too difficult for most people to
00:13:53
navigate and certainly too expensive I
00:13:55
know I'm a lawyer saying that certainly
00:13:57
I think too expensive for most people to
00:13:59
navigate let me know what you think
00:14:00
please do subscribe to the channel
00:14:02
support me that way it is free to
00:14:03
subscribe and as always thank you for
00:14:05
watching
