Joint Committee on the Draft Online Safety Bill

[bridgetroll]

copied from
https://committees.parliament.uk/oralevidence/2714/pdf/

o. 2 Heard in Public Questions 52 - 68
Witnesses
I: Clare Pelham, Chief Executive, Epilepsy Society; Matt Harrison, Public Affairs
and Parliamentary Manager, Royal Mencap Society; Nina Jankowicz, Author and
Global Fellow, Wilson Center; Izzy Wick, Director of UK Policy, 5Rights
Foundation; Ian Russell, Chair of Trustees, Molly Rose Foundation.
USE OF THE TRANSCRIPT
1. This is an uncorrected transcript of evidence taken in public and webcast
on www.parliamentlive.tv.
2. Any public use of, or reference to, the contents should make clear that
neither Members nor witnesses have had the opportunity to correct the
record. If in doubt as to the propriety of using the transcript, please
contact the Clerk of the Committee.
3. Members and witnesses are asked to send corrections to the Clerk of the
Committee within 14 days of receipt.

[bridgetroll]

Examination of Witnesses
Clare Pelham, Matt Harrison, Nina Jankowicz, Izzy Wick and Ian Russell.
Q52 The Chair: Good afternoon and welcome to this further evidence session
of the Joint Committee on the Draft Online Safety Bill. This session will be
examining the difficult subject of harmful online content and how we can
protect people from it. During our proceedings, we may be discussing the
nature of this content in some detail. I advise everyone of this in the
interests of their safety and emotional well-being, including those who
may be watching our proceedings online. Please remember that, if you
are affected by anything we discuss, you can access free 24/7 emotional
support from the Samaritans. You can do this by telephoning 116 123
and by email at [email protected]. Many of the organisations giving
evidence today also have help and support on their websites.
I would like to welcome all the witnesses for today’s evidence session
and, in particular, to address Ian Russell, chief executive of the Molly
Rose Foundation. Obviously there are very particular circumstances that
have led to the creation of the Molly Rose Foundation and led Ian to be
an advocate on these issues. He is also constrained in what he can say
today, in this evidence session, which he will explain to us. Ian, I know
you have a short statement to the committee.
Ian Russell: Thank you. My youngest daughter died on 21 November
2017 and the inquest into Molly’s death is still ongoing. I would like to
thank Senior Coroner Andrew Walker for the diligence he has shown in
his handling of her case.
From what I have heard, there is widespread hope that, for the sake of
other young and vulnerable people, we all may learn as much as possible
from the inquest. However, while the inquest continues, it would not be
right for me to comment about it today, or indeed comment about
anything that I have learned as a result of the inquest process, whether
or not it is already in the public domain.
It is on that basis that I will be making contributions today. The evidence
I will provide will be drawn from the work of the Molly Rose Foundation, a
charity, as Damian has said, that we established after losing our
adorable, seemingly forward-looking Molly. To summarise our charitable
purposes, we aim to connect young people under the age of 25 who are
suffering from depression or other mental illness that puts them at risk of
suicide to the help and support they require.
Despite speaking only about what I have learned from the Molly Rose
Foundation’s work in this area, I believe the charity’s conversations with
other bereaved parents and our connections with many organisations that
work to help prevent the suicide of young people will provide the
committee with useful evidence for this vital legislation. I thank you for
the opportunity to participate in this session.
The Chair: On behalf of everyone involved in today’s hearing, I would
like to express my utmost sympathy for the terrible loss you and your

[bridgetroll]

2
family have experienced. I say that particularly as a father of a 14-yearold daughter myself. We look forward to hearing from you about the vital
work of the foundation and your views on how we can make the internet
a safer place for young people. We are very grateful to you and to all our
witnesses for giving evidence to us today. As you set out, we have
discussed with the members as well the constraints on the evidence you
can give. We are grateful for you making the time to be with us here
today in any case.
Before we start the questions, I know we have a declaration of interest
from Baroness Kidron.
Baroness Kidron: I am chair of 5Rights Foundation, represented here by
Izzy Wick. Also, I have no formal connection with the Molly Rose
Foundation, but Ian and I have worked on many platforms before on
behalf of children.
Q53 The Chair: Are there any other declarations that members may wish to
make? In that case, we will start with the hearing. I would like to ask you
all what your experience is of raising issues of harmful content, content
that causes harm, with the tech companies. The premise of this Bill in
many ways is that the existing complaints procedures do not work in the
way they should and that the companies are not effective enough at
moderating harmful content. If the procedures and processes were
effective, we would not need to legislate to make it happen.
There is a deficiency here and I would be interested in your experience of
working with the tech companies. Do you agree with the Government in
the UK that we now require legislation to create a statutory process for
regulating harmful content online?
Ian Russell: It is our experience that there is frustratingly limited
success when harmful content is requested to be removed by the
platforms, particularly in terms of self-harm and suicidal content. This is
particularly stressful for families and friends who are bereaved by suicide.
It seems to be only when either news stories break in a particularly public
way, or perhaps regulations change, that the platforms respond. Perhaps
the recent 2 September age-appropriate design code end of transition
period is an example. In the run-up to that, there was a slew of changes
of platform regulation that were, coincidentally or not, published at that
time.
It has become our view, and increasingly so, that the corporate culture at
these platforms needs to change. They need to be proactive, rather than
reactive. After all, they have the resources and skills to do this, but it is
so often done as an afterthought. They should live up to their words
about taking online safety seriously and wanting to make their platforms
safer.
Izzy Wick: Thank you for inviting me to give evidence today on behalf of
the 5Rights Foundation. I agree with everything Ian just said. I would like
to point out to the committee that it is astonishing how content that

[bridgetroll]

infringes IP law is taken down within a matter of seconds or minutes.
Compare that to content featuring extreme self-harm and suicide
promotion; it can be a matter of days or weeks before it is responded to.
At 5Rights, we do not issue complaints directly to tech companies, but we
hear regularly from parents and children about their efforts to get in
touch with tech companies, and they are simply not responded to. What
is really needed in this Bill is a mechanism by which Ofcom can
investigate on behalf of children and represent individual children, as well
as vulnerable groups, to advocate for their needs.
Could I make a broader point about content moderation, complaints
processes and take-down? If we shift the focus up stream to address the
systems and processes, we do not get into a place where we rely solely
on content moderation. Let us not forget that content is uploaded in its
billions every minute of every single day. The job of these companies to
police that content is totally disproportionate when we could be focusing
on those very systems and processes, the algorithms, the
recommendation systems, the design features that create this risk, and
spread and amplify harmful content and activity.
Nina Jankowicz: Thank you to this committee for your hard work. It is
heartening to see these issues being addressed on this side of the
Atlantic. In January, at the Wilson Center, we released a report called
Malign Creativity: How Gender, Sex, and Lies are Weaponized Against
Women Online. It is called Malign Creativity, because we found that
abusers online, not just of women but of many marginalised and
intersectional identities, are able to get around the existing rules and
terms of service in order to continue to abuse their targets.
The social media platforms in many cases are aware of these types of
workarounds, for instance using a picture that is poorly edited, changing
the hash of that image by cropping it or changing it to a different colour,
or perhaps making it into a gif, a moving image, in order to evade
detection by the social media companies. They know that this is
happening, but rather than create that hash-tracking database that we
have seen with terrorist content, for instance, they seem to let content
that is attacking women in particular slide.
I have heard in my work, through focus groups and interviews with
journalists, academics, politicians and other women in public life, as well
as normal women trying to make their voices heard, that technology
companies are not doing their due diligence. We receive rape threats; we
are told that these do not constitute a violation of terms of service. I
myself have received threats that I would be taken care of in the streets
or thrown into Guantanamo Bay, but that does not appear to be targeted
harassment, according to these social platforms.
Honestly, it is a huge outlay of time and energy for people who are on
the receiving end of abuse to report this stuff, to do the work of the
platforms, these multibillion-dollar corporations, to report it, track it,
organise it and make sure that, if we want to bring a legal case, we have

[bridgetroll]

4
that evidence. It can be retraumatising and I am, frankly, tired of doing
the work of the platforms for them. There has been a little movement
recently, as folks like me have made more noise about this. Platforms
have started to commit to more help on behalf of women in particular,
but I have not seen a lot of material change.
The most important thing that can be done, which is addressed in this Bill
through the Ofcom transparency reports, is reporting how many reports
the platforms are receiving and what they are doing about them. Often, I
report things that I never even hear back about, so that is a great
starting point to understand exactly where this is going and what the
platforms are doing about it. Are they taking any action? I have seen
tonnes of content directed toward me that I know is not being taken care
of, and I am one person. This experience, unfortunately, is very common
for women in public life. It is simply the cost of doing business in today’s
internet environment. This Bill can absolutely address this lack of due
diligence by the platforms.
Matt Harrison: Thank you for inviting us to give evidence today. We also
welcome the Bill and think it is very much needed at this time. We have
discussions with the social media platforms and they have been quite
productive in those meetings. As witnesses have said before, that does
not really translate to the day-to-day experience of users of those
platforms.
A particular issue that we see is an inconsistent application of moderation
of content. You could have a particular phrase or word used in one
context and it is reported and deleted, but in another context, or in a
slightly different post, tweet or whatever you want to say, it is allowed
under the terms of service. I hope the Bill can tackle those
inconsistencies and bring a standardised approach to how we deal with
these online harms, especially for people who perhaps require a bit more
assistance or find the process of trying to report or deal with social media
companies incredibly difficult, inaccessible and quite stressful at times.
Clare Pelham: I would like to take a few seconds to say in plain
language what I think you mean by online harms and what that means
for people with epilepsy. People with epilepsy—that is 600,000 people,
one in 100 of us, in the UK—are regularly targeted on social media with
flashing images, which may, as I am sure the committee knows, cause a
seizure. That is not just nothing. That means that you can be standing in
the coffee shop, glancing at your phone, with a scalding cup of hot coffee
in your hand, and you can have a seizure. These are targeted attacks.
They are deliberate and hate-filled crimes and we wish the Bill to make
them explicitly so.
In terms of trying to raise this issue with the social media companies, in
May 2020 the Epilepsy Society was targeted with hundreds and hundreds
of these attacks. These criminals, as I would call them, tend to focus on
social media users who use the word “epilepsy” in their posts or talk
about seizures in their messaging. We have needed to employ one person
full time for the last 18 months on reporting and taking down these

[bridgetroll]

tide with them.
The Chair: Clare, do the social media companies recognise the harm that
can be caused by this content being shared? Do they recognise this as
content that they should be proactively looking to act against and close
down the accounts of people who are persistent offenders?
Clare Pelham: I would like to be fair and say that, before this attack
happened in May last year, they possibly did not. It is not well known.
Many people are genuinely horrified to find how frequent it is, but there is
not a scintilla of doubt now that they know. We are in contact with every
platform, if not every day, certainly every week. We have given regular
interviews on every news broadcaster, on television and on radio. There
is absolutely no doubt.
The response, when it comes, tends to be ponderous. Every time this
image is viewed, there is a potential threat to the life and livelihood of
the person with epilepsy who views it. To give one small example, if you
have epilepsy and you work in a nearby market town, should you have
one seizure as a result of seeing a flashing image on social media, you
lose your driving licence for a year. That might mean that you lose your
job and your livelihood. This is not nothing. Even if you do not have a
catastrophic accident, scald yourself, fall over, break your bones, hurt
your head and give yourself concussion, you might lose your job. We feel
that we cannot hesitate in taking down these images.
The Chair: So I am clear, in your experience how are these images being
targeted at people with epilepsy?
Clare Pelham: They tend to be the work of concerted gangs. We know
this because, when they see the result of their flashing images, they send
each other celebratory messages, which say, “One down, so many more
to go”—that sort of thing. This is absolutely ghastly to say. I would like to
say that the Epilepsy Society has a helpline, and we are more than happy
to talk to anyone who is upset by hearing this. They focus on key words.
They are obviously using some sort of patterning device that enables
them to focus on words such as “epilepsy”, “seizure”, “epilepsy meds”, in
order to target vulnerable users.
The Chair: The reason I ask is that, on certain platforms, I would
imagine you might need to use some of the systems the platform has
created itself, or makes available to advertisers, to target people in that

[bridgetroll]

and wide ranging. They can be extreme, from exposure to self-harm and
suicide content, violent sexual pornography and unsolicited contact with
adults they do not know, right the way through to more insidious harms
that might build up over time—the constant nudges and pushes to
engage in certain behaviours, or microtransactions and in-game
purchases nudging them to buy loot boxes. There are also things such as
algorithmic bias and age-inappropriate advertising.
There is no time this afternoon to cover the full range of harms that
children experience, but I would like to stress that they are ubiquitous
and systemic. I am not talking about a few platforms, a handful of bad
actors or even a few pieces of content that have slipped through the
moderation net. The scale at which children experience harm online is
because of the systems and processes that are designed to extend
engagement in order to maximise reach and activity at any cost,
including the safety and well-being of children.
By systems and processes I mean things like hovering over an image, the
algorithms, the recommendation systems, or even just clicking on
something that then determines what you see on your For You page or
your homepage. These are all designed to meet commercial objectives.
I have brought with me today for each committee member a copy of our
recent Pathways report, which illustrates those pathways to harm and
exactly how those automated processes lead to quite extreme risks for
children. I would encourage each committee member to look from page
16 onwards, which I should warn features some very distressing images,
but they illustrate just how quickly these accounts can be inundated with
a deluge of harmful content and activity.

[bridgetroll]

7
There is a 15-year-old girl who, within days of setting up her account, is
sent deep into a rabbit hole of thinspiration and unrealistic body ideals.
There is a 14-year-old boy who is sent numerous sexually explicit
messages from unknown adults. These are just snapshots. We know from
DCMS’s own research that 80% of six to 12 year-olds have experienced
some kind of harmful content online. Half of 13 to 17 year-olds believe
they have seen something in the last three months that constitutes illegal
content.
No environment can be 100% risk free. We want to make sure that,
rather than creating a walled garden for children, we are doing
everything we can and taking every available opportunity to prevent risk.
If the Bill is going to deliver for children, it needs to focus on those
systems and processes that create risk.
Q55 The Chair: Finally from me to Nina, last week the committee discussed
racist abuse online, particularly in the context of football. One of the
issues we discovered there was the extent to which hate speech is
normalised. Do you think there is a growing problem of the hate speech
directed towards women, outright misogyny, becoming normalised
through its use on social media and becoming something that, as things
currently stand, people are basically told they have to accept if they have
any kind of public or prominent position? It is obviously totally
unacceptable, but it seems to be something that no real control is being
brought over.
Nina Jankowicz: Being a woman online is an inherently dangerous act.
That is the long and short of it. It does not matter what you do. You are
opening yourself up to criticism from every angle. I would like to say that
this online misogyny is being hastened by social media. I think people
feel that they can say more vitriolic things on social media than in public,
but the truth is, unfortunately, that misogyny is not new.
These abusive messages and this abusive content are not necessarily
going to stop. It is silly to think that we could turn the dial down to zero,
but we can change the systems, as Izzy was just saying, so that targets
are not just hit with this stuff over and over when they are being
dogpiled, and repeated abusers who are engaging in this sort of
behaviour all the time are no longer welcome on those platforms. Those
are very workable solutions that still also protect freedom of speech,
because they are grounded in the terms of service that the platforms
themselves are supposed to be implementing.
Online misogyny is growing and it is something that, unfortunately, any
woman who wants to be in any sort of public position has to endure right
now. Many women are changing what they write, what they speak about,
what careers they choose to pursue because of that understanding that it
is part and parcel of existing as a woman on the internet.
The Chair: In this context, the question of freedom of speech would
appear to be less about speaking truth to power than one person’s
perceived right to abuse another.
8
Nina Jankowicz: Yes, exactly. There is an analogy that I often draw,
because it is difficult for people who have not experienced online abuse to
understand the effect that it has on a person. Imagine that you are
sitting at your computer, whether that is in your home office or here in
Westminster, and your office is filled with people who are shouting filth at
you, dissecting every part of your appearance, reducing your résumé and
all your hard work to your gender, saying that you deserve to be
tortured, raped or worse, and you are expected to continue to do your
work.
If I were to listen to people’s advice and ignore the trolls or simply turn
off my Twitter account, I would not be in front of you today. The internet
is how I do my work. The idea here is to quash women’s right to freedom
of expression. It is to take them out of the public eye, because women,
according to the purveyors of this abuse, do not deserve to be there. We
need to stand up to that. We need to protect that right to work, that right
to freedom of expression. That is at the core of this misogynistic abuse.
Q56 Debbie Abrahams MP: Good afternoon, everyone. I wanted to ask you
what you thought the relationship is between individual harm and societal
harm.
Nina Jankowicz: This is something I think about a lot, because we hear
from abusers of women online that we are just being too sensitive: “If
you can’t take the heat, get out of the kitchen”. That has been said to me
over and over. The issue at hand is a democratic issue. Let us take VicePresident Kamala Harris as an example. In our report, over a period of
two months ahead of the US election, we gathered 336,000 pieces of
gender disinformation and abuse about 12 candidates for office in the US,
the UK, New Zealand and Canada; 78% of that was directed at then
candidate for Vice-President Harris.
I think of the little girls who were so happy on inauguration day, seeing
our first female Vice-President inaugurated and, for those of them who
are online, looking at the responses to tweets or Vice-President Harris’s
Instagram posts, where people call the Biden-Harris Administration “Joe
and the ho”, or worse. There is much worse, alleging that Vice-President
Harris slept her way to the top. That is going to change how young
women engage in the democratic process, whether that is running for
office or simply engaging in democracy writ large. That is one concern.
There is also a national security concern. In the report, we do a deep dive
on three journalists who have been targeted by Russia, China and Iran,
which recognise this endemic misogyny in our societies and use it to
undermine our democratic society. It is not just a democratic concern; it
is a national security concern, which should make it of interest to
everybody in government. It is not just about hurt feelings. It really
affects the way our countries operate.
Clare Pelham: It seems like the right moment to ask that question, as
we have just reached the end of the Paralympics. The analogy that I
would draw is that, as a country, we have done so much to integrate
9
disabled people into our national life and to increase accessibility in so
many ways. The Paralympics is just one example of that, Disability
Confident another.
The use of social media to attack disabled people is undermining all that
good work. It is pushing disabled people back into the margins of society.
It is excluding them from the support that would otherwise be available
to them and from the forum that is, very naturally, accessible to them.
We are all very cognisant of the fact that physically disabled people may
have difficulties of physical access. I think we all understand ramps and
other sorts of infrastructural support that we need to offer people with
physical disabilities. The awful thing about the criminals who are
targeting people with epilepsy is that they are now excluding them from
the medium that is most accessible to them: online communications. We
have seen people targeted who have said, “I’m newly diagnosed with
epilepsy. I’m looking for support”. The support they get is to be
inundated with flashing images that might give them a seizure. That is
the link I would make with the societal issue and the individual issue
here.
Izzy Wick: One in three internet users around the world, and one in five
in the UK, are children. That is one in five with all the vulnerabilities
associated with their age, one in five who are routinely ignored by
platforms that do not respect their rights and needs. The silencing effect
on children of a toxic digital environment has real-world societal
implications. We know that teenage girls are much less likely to speak up
on social media for fear of being criticised. That has a very real effect on
their ambitions to go into public life. There is a very clear link between
individual and societal harms.
Matt Harrison: I would echo some of the comments. There is definitely
that direct link between individual harms and societal harms. I guess
there are two angles to that. First, in terms of learning disability, there is
that reinforcing of negative attitudes and stigma towards this group. Over
the last few years, this has changed in a positive direction, which has
been very welcome. That very vocal element on social media platforms
that, to a degree, normalises discriminatory words, very harmful words
and phrases, starts to unravel those threads of the work that lots of
people with learning disabilities themselves have been doing on social
media.
We are particularly concerned about young people, both with and without
a learning disability, who see this content and think that it is normal.
They are obviously very impressionable, so that moves through into their
adult lives.
Then there is the impact on other people with disabilities, who see this
content or hear of friends who are subjected to it. It tends to reinforce
these exclusionary feelings and social isolation, which are already issues
for people with learning disabilities, especially for young people. We know
that on average one in three spends less than one hour a day on a

[bridgetroll]

10
Saturday outside. They are very socially excluded, and social media is
one avenue for them to connect with communities, but, once they start
seeing these online harms, this harmful content, it normalises that
behaviour. It really knocks people’s confidence and has a damaging effect
on their life chances.
Debbie Abrahams MP: What do you say to those people who say,
“We’ve always had misogyny and racism in society. Propaganda,
misinformation and disinformation have been around for hundreds of
years as well”? What makes the difference between an analogue and a
digital age? This is not just about the speed at which it can spread. Is
there something special about what happens to people when they use
platforms?
Nina Jankowicz: The biggest change is the quantity; it is not just the
speed. If you visualise what abuse against a person looks like online, it is
like you are a tiny dot in a swarm of insects. That is what it feels like.
Your phone is buzzing. It is getting hot. You cannot look away, because
you are trying to control the situation, which is impossible, of course.
That causes very real psychological harm. It is very different. Although it
would be terrifying to have someone shouting these things at you on the
street, you would call a police officer and they would take this person
away or enforce some space. Online, no one is doing that. That is the
biggest difference.
Matt Harrison: You spoke about speed, which is an issue. The dogpiling,
piling-in incidents are very different from what perhaps a similar harm
would be in person. Perhaps it is one person, or a very small group, who
is probably known to that individual. Online, there is that element of
speed. You start drawing in people from around the country. You feel like
you are being attacked and on a much larger scale perhaps than in the
real world.
It is about the ease as well with which people can carry out online harms.
Perhaps there is a feeling of invincibility in posting these online harms,
which you may not have in real life, in the sense that seeing someone
face to face might make you think twice about using a particular phrase
or a word. Online, you feel that distance between you and—I would not
say your target—the person receiving that messaging, in some cases.
Clare Pelham: It is actually, and perhaps counterintuitively, far worse
than having this happen in real life. If I were walking down Hammersmith
high street, and somebody came up to me and pushed me over or
punched me in the face, I would be immediately surrounded by wellmeaning people looking to help me up, who would call a police officer and
run after the person who had assaulted me. The awful thing when this
happens to you on social media is the feeling of isolation, that you are
completely beyond the reach of the law. You are in some sort of badlands
where there is no criminal law, no protection, no possibility of
prosecution.
11
Even worse than that, you are cut off from the support of others in your
community, because of course what happens is a terrible fear. You are
targeted by these wicked people and anyone who reaches out to help you
is themselves targeted. This has happened to us so many times. We try
to protect someone with epilepsy who is targeted—we have numerous
examples, some of which we have provided to the committee—and the
charity is targeted. Once, the BBC very kindly retweeted one of our
messages. The BBC was targeted. It is pervasive and isolating. You feel
completely powerless.
Q57 Lord Knight of Weymouth: There are a couple of things I wanted to
follow up, particularly starting with you, Nina. You talked about the
possibility of repeated abusers no longer being welcome on platforms,
which is something we would all want to achieve. Is it feasible for us to
end up with some kind of a system so that, if a bad actor is identified on
one platform, other platforms then prevent that person being welcome?
Nina Jankowicz: I would like to see something like that developed. In
the report, we recommend the development of something akin to the
Global Internet Forum to Counter Terrorism—GIFCT—which is the hashtracking database I mentioned before. Whenever a piece of terrorist
visual content is shared, its individual hash is shared among many
platforms, not just the large ones. It deals with photo-sharing services as
well that are quite small.
There is this desire among social media platforms to address that issue,
but, when it comes to the issues we are all talking about today, we have
not seen that alacrity. Something like that could certainly be used. As an
example, going back to Vice-President Harris, we saw so many images
shared over and over in the campaign period. Using Facebook’s own
CrowdTangle social listening platform, we were able to do a reverse
image search on the platform and draw up thousands and thousands of
hits for that image. Only after a Washington Post article was published,
talking about this phenomenon online, did Facebook take them down.
So it is possible. They have the technical capacity to identify this stuff.
Sharing it across platform would be extremely effective. In the report, we
saw those narratives jump from the less mainstream or more alternative
platforms such as Reddit—Reddit is one of the most popular online
platforms, but it receives a lot less scrutiny—8kun and Parler, where a lot
of the worst abuse is, to the more mainstream platforms, where they are
much more effective. We found that abusers like to yell at their targets,
not just about them. That sort of sharing would be extremely useful,
particularly for visual content.
Lord Knight of Weymouth: You mentioned the terrible abuse that has
been directed at Kamala Harris. The Bill as it is currently constructed has
content of democratic importance being essentially excluded. What is
your view on that? Is it a loophole? I am interested in all the panellists’
view on that, as well as journalistic content and advertising, to an extent,
as to whether we have loopholes here that we need to address.
12
Nina Jankowicz: Those are actually two of the things that I had a little
bit of pause about in the legislation, because I think there are many
citizen bloggers who would view themselves as journalists and would like
to be included in those exceptions under the law. Certainly some of the
people who have targeted me think of themselves as journalists. Many
people would say, “My freedom of expression is being quashed by this
content being removed during an election season”.
In that case, we have to go back to the platform’s own terms of service.
Gendered abuse, targeted harassment and racial abuse all go against the
rules the platforms themselves have set out. That is where I go back to
the transparency reporting and asking, “How many reports are you
getting about this content and what are you doing about it?” Facebook of
course likes to say, “90% of the content that we find we take down”, or
something like that. I am just throwing that out and making that up, but
it likes to self-congratulate in that way. It is not finding a lot of this
content, as we found.
There are loopholes there, but, in that case, we have to go back to the
platforms’ own standards that they are setting out themselves.
Lord Knight of Weymouth: Izzy, your excellent written evidence
includes lots of amendments, except on this area, where you have just
said that we perhaps need to think about the definition a bit more.
Izzy Wick: We are concerned about the definitions and these carve-outs
for content of democratic importance and journalistic content, mainly
because of how they have been drafted. It is very difficult to see how any
type of content could not reasonably be considered journalistic content,
or indeed that any user could not simply say, “I’m a journalist. Therefore,
this platform shouldn’t be taking down my content”. Our concern is that it
creates a loophole and will undermine the very good safety objectives
that these services will have to comply with.
You mentioned advertising. The exemption for paid-for advertising takes
out of scope a huge category of potentially harmful content. A big
concern is that it also does not match up to existing regulation that we
have under the video-sharing platform regime. This regulation imposes
certain requirements on platforms that feature paid-for advertising that is
directed at children. It seems crazy that the Online Safety Bill, which will
supersede that regulation, will weaken those requirements. We think the
Bill is an opportunity to bring all online advertising regulation on to a
statutory footing. I urge the committee to look in particular at that
exemption.
Ian Russell: These exceptions, as well as potentially providing a
loophole, could also undermine public confidence. That is one thing that is
really important in this Bill. The public are looking. The parents I meet
who have concerns about their children—and all parents have concerns
about their children—want to know that something is being done to help
them, because it is a big job. There is no parent who can do that by
themselves, and they are looking to this. If anything were to undermine
13
the confidence in this Bill, that would be a tragedy, because this is a
chance to do something to help them and help the country.
Matt Harrison: I fully agree. It is quite a dangerous loophole actually,
especially given the growing importance of social media platforms in
journalistic content. I am sure we all follow many of the journalists who
cover Westminster. This shows you that their following has exploded over
the last few years. We are quite concerned that there is a loophole.
Preparing for today, I was really struggling to think of an outside scenario
where an online harm or a discriminatory remark against disabled people
or people with learning disabilities would be in the public’s interest or of
democratic value. It is a loophole and an interesting question, because
there are a couple of phrases in the legislation that are very ambiguous.
They are open to interpretation from various angles and could undermine
the intentions of the Bill.
Clare Pelham: I wonder if a way of looking at this issue analytically
might be to compare it with criminal offence detection, because what we
are talking about here is hate crime. Revealing somebody’s true identity
on social media is a deterrent, so people have to take ownership of what
they are saying and doing, and take responsibility for it. You can use
technical devices to prevent material containing certain words or, in the
case of epilepsy, certain images being disseminated. That is preventive.
Ultimately, you can use algorithms for speed of removal. That is after the
fact. You have those three phases of prevention, detection and conviction
there, because this is not just behaviour; it is criminal behaviour, and it
helps sometimes to think of it in that way.
Q58 Baroness Kidron: Ian, I want to pick up first on something you said.
You said that parents are hoping that the Bill will do something for them.
In many conversations that I have, people ask, “Can’t parents look after
children?” I wonder whether you can speak to the committee about what
it feels like to be a parent, what the parents are saying to you about what
they are finding their kids doing, and how realistic that is for parents.
Sorry, that was not very articulate.
Ian Russell: It is fine; it gave me time to think. It is a tough question,
and it is one that I struggle to answer, because it is so dependent on
circumstances. Some parents are concerned with what their children are
seeing online and they say, “Should I ban my child from having a
phone?” My reaction would tend to be, “That is an overreaction”, but of
course there would be some circumstances where that would be entirely
the right thing to do. You cannot give a straightforward answer, even in a
fairly simple case.
To give generalisations is hard, but a concern expressed is that, just as
the parents catch up with Facebook, then Instagram, and then Snapchat,
along comes TikTok and many other platforms. They do not have time to
look into all the safety implications of all those platforms. Even if they
did, their children would be one step ahead of them.
14
If you widen that beyond those big and known platforms, I have talked to
families who have lost children to suicide, and platforms such as Discord,
Roblox and Wattpad have been included among them. These are writing,
communication or gaming platforms. In fact, none of these platforms is
designed to do harm. It is the fact that they can be used to do harm and
be misused in that way.
Picking just one of those at random, because it was something I
discovered this morning—I was doing a little bit of research—on 8
February 2020, the Times reported that a search returned 197,000
results from Wattpad targeted at suicide, for example. That is a writing
platform. It shows how a seemingly innocent platform can be used. If you
are a parent facing the multiplicity of platforms and the fact that the
platforms you know about will be different from those of your children,
you will want to know how you can have help in answering that and in
supporting your children to use the internet safely.
Baroness Kidron: Can I pick up on that? I did not get the number, but
that was a lot of returns from the search. Do you think that we are
missing the point when we think about content as being the content that
one bad actor sends to another, or one bad actor sends to a child?
Actually, there is this swathe of availability.
Ian Russell: There is a swathe of availability: 197,000. That is a big
number. There are different sorts of orders of magnitude. Direct online
bullying is particularly bad. On a trip to the US, I met another parent
bereaved by suicide and their child had been online-bullied directly, direct
messaged by people at their high school in New Jersey. Their story was
tragic and horrible. Because the bullying from a small number of people
was so direct and so targeted, it probably had a greater effect on that
child.
The huge amount, hundreds of thousands, of posts of content that can be
harmful in the very many other ways that harm can be brought to bear,
particularly in young and vulnerable people, means that, just by going
online, you are dipping your toe into dangerous water. The beach might
look shallow and quite calm, but it is all too soon before you are sucked
out of your depth, if you like. Just a few misplaced search terms can send
you in the wrong place.
In a way, it is being a person online that is dangerous. We all have to
admit that. It does not matter what age you are. We have to remember
that, every time we go online, there is potential danger. We have to
moderate our behaviour online and think about how we are affected when
we go online. It is hard enough for an adult to do that; it is very hard for
a young child or a vulnerable person to do that.
Baroness Kidron: I am going to pick that up and ask Matt a question
that others may want to come in on. I do not think there is anybody in
the room who does not think that some of the services and products are
valuable. In particular, the reason I am focused on you, Matt, is this. I
remember a young man who told me that he was getting mental health
15
support online. Then he read, retrospectively, the data and privacy
agreement and he stopped going to sessions because he did not feel safe.
I guess this is my question to you: what is the opportunity here, as we
talk about harm, to provide really excellent and trusted services? Is that
something that we are missing in the Bill, as far as you are aware of it?
Matt Harrison: That is an interesting question. I think we all see the
benefits that social media can have for people. I do not think anyone
actually doubts that, but I can understand where you are coming from.
The main point of the legislation at the moment, or the need for it, is to
tackle the negative, harmful content on there. As you said, there are
quite a lot of good things happening on social media. Social media
platforms tend to be quite good at promoting positive content, for want of
a better word.
The point is that legislation is mainly about tackling online harms, but
there is a key opportunity within the legislation, under the extension of
Ofcom’s media literacy powers, to do some of what you were saying
about empowering people to use social media better, to stay safe online,
to tackle online harms at an individual level perhaps, or at least to
empower others. Children and young people especially perhaps will
benefit from Ofcom’s education down the line, as they have with other
bits of media.
I have gone around the question, but there is a role for promoting good
content. I do not think that is what needs to be tackled in this Bill. It is
more about tackling the negative side of social media.
Izzy Wick: It is important to say that the digital world is not optional for
children and young people. We want children to be engaged in the digital
world and to flourish online. This Bill needs to set the expectations for
companies and establish rules of the road. We need Ofcom to produce
minimum standards on very basic things such as terms and conditions,
reporting and redress mechanisms, and age assurance. There should be a
requirement in the Bill for Ofcom to do that, which will then establish the
floor of protection.
To pick up on Matt’s point about the role of education, I do not think
children or their parents should be expected to mitigate risks that need to
be addressed at the level of system design. We cannot expect parents to
babysit the technology as well as their children. Also, no parent can be
with their child 24/7. As Ian said, parents often feel their children are
much more digitally adept than they are.
To make a broader point about the role of education and digital literacy,
we should not be educating children to navigate pathways to harm or
inherently unsafe digital spaces that do not recognise their rights and
needs. I would also like to point out that many of the existing digital
literacy programmes are either provided or funded by tech companies.
Perhaps unsurprisingly, they focus much more on user behaviour, on
topics such as stranger danger and bullying, and much less on the risks
that are built into the fabric of their systems design. We need Ofcom to
16
establish minimum standards for any education initiatives that are
designed for children.
Baroness Kidron: Nina, I am seeing a lot of nodding there. I am
interested in minimum standards. I heard your answer before: “It’s
difficult to know where a line is”. Do minimum standards that are then
upheld—you cannot celebrate misogyny on a site; you have to take some
action—work for you?
Nina Jankowicz: Yes. The minimum standard is, theoretically, already
established in the terms of service of many of the large platforms. If you
leave aside the alternative ones that are gung ho—freedom of expression
and fairy dust everywhere—for Ofcom to be able to establish the
minimum standards that would be applied to all platforms and incur fines
would be a useful starting point. That could be based, again, on the preexisting terms of service. Theoretically, these stipulations already exist.
They are just not being upheld.
Clare Pelham: I would like to say thank you for reminding us that lots
and lots of good things happen on social media. Lots and lots of us have
fun times being on social media. To draw an analogy, if social media is a
wonderful playground, we do not need Twitter or Facebook to tell us how
to play on the slides and swings. We need them to stop bad people from
coming into the children’s playground. We know how to have good fun on
social media. We just need the social media companies to stop this
wonderful, joyful opportunity that we all have to communicate with each
other being abused by people with bad intentions.
Q59 Lord Clement-Jones: I want to take this forward on the things that
need to be done on top of what is already in the Bill, looking beyond what
Baroness Kidron has just been teasing out, and starting with you, Izzy,
on the point about the mandatory nature of the child online safety code
and expanding that into other codes.
You have gone so far as to actually put out amendments. You are making
a distinction between what Ofcom recommends and what it is absolutely
mandating. I am wondering whether there is a real substance behind
that. You could argue that Ofcom will expect those standards to be
maintained.
In particular, you then go on to talk about the question of risk
assessment, which is clearly crucial. Ofcom itself is very keen on risk
assessment. It talks about it in the AVMS context and so forth, even
though it does not have the statutory power to insist on it. Clearly, given
half a chance, it is going to want to go forward with that. You have a
particular four Cs way of looking at risk assessment. I wonder if you
could unpack those two thoughts and then see what the others have to
make of it.
Izzy Wick: I will start with your point about codes of practice. It is worth
saying initially that we have statutory codes in other areas of digital
regulation. If pro-competition statutory codes of conduct can be made for
firms with strategic market status, they can absolutely be made for child
17
safety. There is no reason why they should not be, but this comes down
to making Ofcom’s life a bit easier.
This is going to be quite difficult legislation to enforce if those
requirements are not set out on a statutory footing. It will also aid
compliance for the companies to say, “Okay, this is the list of things we
have to do”. To what you said about risk assessments, we are thrilled
that there is a requirement in the Bill for services to conduct child risk
assessments. Unless the content and process for assessing risk is set out
in an enforceable framework, we will be leaving these companies to mark
their own homework.
Lord Clement-Jones: I saw Matt nodding vigorously at that point.
Matt Harrison: Yes, it is a very good point. I was alluding to that before
perhaps with regard to the ambiguities in the Bill. The previous panel was
asked a question about protected characteristics being listed in the Bill. It
is a slightly different point, but that is something we would very much
like to see. That has a flow-through effect in terms of other elements of
the Bill. Once you start explicitly enshrining those protected
characteristics, it gives the Government and Ofcom a sense of direction,
or at least a clearer sense, of how and where they can enforce. We would
very much like to see that.
I fully agree about the risk assessments and marking their own
homework. Social media companies are obviously incredibly skilful at
developing their internal processes of moderation, algorithms and AI. It is
mind blowing to me. There is a real challenge for regulators to
understand and get their heads around this and therefore understand risk
assessments, which can be written in incredibly technical language and
sound impressive on paper but, in reality, may have very little impact on
day-to-day users.
Lord Clement-Jones: You are almost saying, “Specify that you must
assess the risk of the algorithm itself” as part of the rubric.
Matt Harrison: Yes. I guess there are two points, aren’t there? There is
the actual risk assessment itself and then there is Ofcom’s ability to
understand what is being put in front of it and the algorithms of
companies that are at the forefront of these very complex systems.
Nina Jankowicz: I will draw an analogy to Germany’s NetzDG law,
which is similar in some ways to what is being proposed. The specificity
issue in Germany has been extremely important and hindered some of
the progress that Germany has made on regulating social media and
content moderation. In their transparency reporting that they have
required of the social media companies, they have created blanket
categories that do not necessarily apply to every platform. Obviously the
types of engagements and types of content on each platform vary
greatly.
18
The more specificity that is asked for, for both reporting and risk
assessment, the better. Those should be tailored to each platform as
well. Comparing YouTube and Facebook, or YouTube and TikTok, against
Parler or whatever is like comparing apples and oranges or a cow to a
tree. They are really quite different. Making it more specific will render
more practical data and give you a better understanding of what is going
on. If you are at 40,000 feet, you are really not going to see what is
going on in the weeds.
Lord Clement-Jones: Do we have the models to do that? We are not
just praying for some great risk assessment that has not yet been
invented.
Nina Jankowicz: The social media platforms can do that if they are
compelled to. The point that Izzy made before about marking your own
homework rings true here. You or Ofcom would need to determine
exactly what measures you would like to see and compel the social media
platforms to hand over that data. They have it. It is just a question of
getting it from them.
Ian Russell: Safety standards, risk assessments and method statements
have all become part of modern business life. They are, quite rightly,
expected, but they are conspicuously absent in the online world. They are
there in places, I am sure, but they are not as prominent. In all this
legislation, it is most important that we mirror online what we live by in
the offline world. That is what is missing at the moment. Sixteen years of
social media and 30 years of the internet has led us to a place that there
is no mirror of the accepted normal standards that apply offline. There
should be equivalence in the online world as well.
Clare Pelham: I will not repeat it, but I would very much like to endorse
Ian’s last point about equivalence. That must be our guiding principle. It
should not be possible to do something online that would be unlawful if
we did it in real life.
The key to securing good regulation is the provisions on personal
responsibility for senior managers in the social media companies. I have
been in a position myself where I was personally responsible and, my
goodness, it focuses your mind. You will work with the regulator so
closely to be so accurate and so well defined in every aspect of the
regulatory detail. It seems to me that, if we can secure that, everything
else will flow from it.
Q60 John Nicolson MP: Mr Harrison, you were talking about language and
why language matters on social media. We heard in our earlier session
about the way in which racist language seems to be filtering from the
internet back to schools. After decades of improvement, we were told
that it is getting worse in schools. Is it the same with the language that is
used about mental health? Are you seeing a knock-on effect in everyday
society?
19
Matt Harrison: That is a very interesting question. Unfortunately, data
on learning disability is very sketchy and hard to come by. We did a
survey, I think in 2007, and found that 82% of children with learning
disabilities had been bullied, which equated at that time to about 280,000
children, give or take. In terms of measuring how it moves from the
online space into the real world, it is incredibly difficult to know, so there
is definitely a data gap there. There must be some sort of filtering across.
John Nicolson MP: I make a point of always writing to anybody I ever
hear misusing the word “schizophrenic”. It is informal, but I think it is
getting a bit better. You would regularly hear on the BBC people saying
that they are schizophrenic because they cannot make up their mind
whether they are for or against Brexit or whether they prefer tea or
coffee. That matters, as you know better than most people, because if
the word “schizo” is used, which is a horrible word, or “schizophrenic” is
misused, it is very difficult for people who are told that they have that
condition to embrace the condition, because who would embrace a
condition that has such horrible connotations?
I will not say who they are, but over the years only two people, one a
politician and the other a journalist, have fought back when I have said,
“Please don’t use that expression”. Most people say, “I’m very sorry. I’ll
never use it again”. Do you think it is becoming worse? I am noticing a
bit on social media that some of those words that have not been used for
a while or have been diminishing in their use are coming back into usage.
Matt Harrison: It is a very good point. There have been perhaps a
couple of high-profile cases of people with learning disabilities being
targeted. I am thinking particularly of Katie Price’s son Harvey. I wonder
if there is a knock-on effect from a high-profile case and this appearing in
people’s feeds. As I referred to before, that is a normalisation of those
words and the untangling of positive influence that we are having. It is
very difficult to understand the whole trend. I would love someone from
any of the platforms to give me the data on that, but you see a greater
prevalence of the use of those words online than in day-to-day life.
John Nicolson MP: Ms Pelham, can I move on to you? I just sensed a
collective shudder among members of the committee and witnesses when
you described what it was like to be sent those flashing images. I confess
I did not know anything about this until you wrote to me and asked me to
write a piece about it for one of the national newspapers. Let us be
honest: you wrote the piece and I just slapped my name on the top and
pretended I had written it, so let us have a bit of candour here. It was a
very good piece and I was very happy to put my name to it. You want
something called Zach’s law put into action. We will talk about that in a
second.
I would like to know more about the motivations of people who do this.
Who are these awful people? You refer to them as gangs. Who would
want to send flashing images to somebody living with epilepsy, which
could cause them to crash their car?

[bridgetroll]

20
Clare Pelham: It is almost impossible to understand. We have really
tried, because we thought that if we could understand we might be able
to work with these people and help them to understand that their
behaviour is causing such pain to the victims of it, but all we can think of
is that they are people who have not yet had the education and exposure
that many of us have had to what an equal society that includes disabled
and non-disabled people—
John Nicolson MP: It is a bit more than that, is it not? You might not
care very much about disabled rights, but you would not want actively to
kill somebody. Are these people psychopaths? Who are they? Who are
these monsters?
Clare Pelham: All of us at the Epilepsy Society have to believe that, if
we work well and effectively enough, we will be able to defeat stigma
against people with epilepsy, but throughout history there has been a
very sad tradition of being unkind and hateful to people with epilepsy. To
pick up the point that you made earlier, this was very much on the
decline until the popularity of online media and it has bounced up. It
really has. I am afraid we have not done a survey about it, but
anecdotally we can see it. There are many people with epilepsy who will
not tell you that they have epilepsy because of these folk out there, and
it is heartbreaking.
John Nicolson MP: Could it be that they just do not think that people
online are real people?
Clare Pelham: There is an element of that. It is a very different thing to
sit in your bedroom doing hate-filled things, perhaps after a drink—I do
not know—than it is to look somebody in the eye. You mentioned a lovely
young man called Zach Eagling, but surely nobody, if they had to look
him in the eye, would send a 10-year-old child with cerebral palsy those
messages. I cannot believe that they would, but they are definitely in
need of some help to understand the consequences of what they do.
John Nicolson MP: What do you want when you talk about Zach’s law?
As you say, he is a lovely wee boy who has been targeted. What do you
want Zach’s law to do? In fact, could we effectively incorporate your
proposals as Zach’s law into this Bill?
Clare Pelham: We earnestly at the Epilepsy Society hope that you
would. We would love you to implement a recommendation made by the
Law Commission that the intentional sending of flashing images to a
person with epilepsy with the intention of causing that person a seizure
should be made a criminal offence. If you could do that in the Bill, it
would unlock so much improved quality of life for people with epilepsy, I
cannot tell you.
John Nicolson MP: I suspect that, if you thought you were going to
serve a term in the slammer at Her Majesty’s pleasure, you might not be
tempted to send flashing images.
21
Clare Pelham: Completely, and, if I may add to that point, so often,
sadly, the way to people’s hearts, especially at large profit-making
companies, is through their wallets. Accompanied with the financial
sanctions that the Bill provides for social media companies, we would see
within weeks the social media companies working to devise algorithms
that would prevent these images being sent at all.
John Nicolson MP: Why do you think the social media companies do not
take this seriously enough? We have heard from all the witnesses today,
and we heard from witnesses in our first session, the same recurring
theme that social media companies do not take things down, unless it is
copyright things. They are very quick with copyright things, but if it is
misogyny, attacks on people with mental health issues, or terrible cruel
attacks on children being bullied, they just do not act. Why is that?
Clare Pelham: We have been through a very long journey of education
and persuasion, and I am really sorry to say this but I am afraid that the
only answer that we are left with is to follow the money. A company that
can act on a breach of copyright that may have a massive financial
impact on its bottom line is surely able to act just as quickly. The reason
it does not is that the “only” consequence of its inaction is that some poor
person gets hurt. I am really sorry to say this, but it feels like the only
explanation that is left. We have had numerous very private
conversations with people who work at these companies, who have asked
us not to use their name, who have said their business model does not
allow them to spend any more money on this. That is just heartbreaking.
John Nicolson MP: Although we know from Germany that, if they are
threatened with a huge fine, they suddenly start hiring people and they
will take down harmful content. You mentioned that Twitter gives you
money to help you cleanse your account. How did that come about?
Clare Pelham: We communicated with them that, as a charity, we were
employing somebody designed to help people with epilepsy to do what
we felt was their job, which is to police their social media platform, and
they once gave us £7,500. Of course it is welcome, but it does not go
anywhere near the amount of fun runs, raffles and cake bakes that we
have had to have to raise the money to pay somebody to do this full
time.
John Nicolson MP: Twitter gave you money to help you make Twitter a
less bad place.
Clare Pelham: Yes, absolutely.
John Nicolson MP: They are paying you to hire somebody to write back
to them and tell them that some of this stuff that is appearing, that is
targeted at you, is horrible.
Clare Pelham: That is it.
John Nicolson MP: You have to pinch yourself, Chair.
22
The Chair: Indeed, it would not even cover an air fare to California for a
tech executive.
John Nicolson MP: Certainly not the way Twitter executives fly
backwards and forwards, it would not. I will leave it at that.
Q61 Lord Gilbert of Panteg: I will pick up there, because what I have in
mind is potentially much more expensive for the platforms. Before I get
into that, we all pretty much agree that tackling these issues is going to
require the bringing together of content policy, systems and design
regulation. I would add competition policy, which is strikingly missing
from this Bill and could be a very important part of it. I just want to pick
up on the competition element.
The nature of these platforms, and certainly the larger platforms, is that
they are very dominant in the public space and, if you are a young
person or, indeed, if you are anybody and you want to engage with your
friends, you need to do it through a very small number of platforms;
otherwise, you will not find them and you will not be able to talk to them.
These platforms have it in their power to set the terms and conditions
and to provide toolboxes that you can use as a user to regulate the
environment in which you operate. We have not gone anywhere near
their business model, and it seems to me that they want this to be a
debate about taking down content and not a debate about their business
model, which leads to this kind of behaviour.
I would be interested to hear from probably all the panellists on this. This
potentially seems to be where competition policy comes in. I cannot see
any legal, technical or other reason why we should not say to platforms,
“You must totally open source your platform. You remain the platform.
You maintain the underlying database, but you work with a variety of
other organisations, so that they can provide toolboxes to their users to
access the platform through their own gateway, be it by way of an app or
an interface”. They might work with you, Clare, to create an interface
that people who are particularly vulnerable in your sphere can use to
access these platforms with very high levels of safeguards.
The parallel, to me, is with the regulation on open banking, which has
worked incredibly well. I access all my bank accounts and financial fees
through one app, which gives me an additional set of tools. I wonder
whether looking at this through a competition approach and saying, “You
are dominant, and you need to share your code and work with other
groups to enable them to provide walled interfaces into your platform”,
would give something really tangible to users to take some control of
their own environment in which they use these platforms. Who wants a
go? Has it been explored?
Izzy Wick: You have hit the nail on the head. There is a very profound
tension between the fundamental business models of these platforms—
platforms that are often provided free of charge, rooted in targeted
advertising and therefore designed to extend engagement, maximise
reach and maximise interaction—and the development of safe and
23
healthy spaces for children online. We need to look at that fundamental
business model at the very root of the problems that we are discussing
here today if we are going to address them in any meaningful way.
The comparison with open banking is a good one. Having greater
transparency, as well as accountability, will do a lot to detoxify this
environment, but what we are also talking about here is enforcement,
and these are companies with incredibly deep pockets. Really, if we want
to see the sea change that is needed across the culture of this sector, we
need to have company director liability, as Clare said.
Ian Russell: It is important to encourage the good. I have not said
enough today how good the internet is. It is a fantastic resource. It is an
amazing thing, and the danger is that all that potential is going to be
spoiled by bad actors who use it. I would have thought it was in the
interests of the tech platforms to be more responsible about how they
provide better safety for their users, as they call them. I am surprised it
has not already happened. That requires a change of corporate
responsibility from the platforms, and that is why I would say that
criminal liability for senior managers is important, because you need to
push hard and quickly with a big impetus to make that change happen.
They have not needed to work in any other way.
Then legislation like this could encourage them and help the tech
platforms to see that they do tremendous good and, if they focus on that
and help the world weed out the bad that is on the world wide web, it is
good for them. That is where competition might come in and help them,
because they will see that it is a way to help their users and to help them
gain a bigger market share.
Nina Jankowicz: There is an example, slightly afield of normal social
media, where this has come up and that is the dating app Bumble, which
is the app where women can message men first. Its lobbying team has
taken it upon itself to pursue some really strict legislation about
unsolicited nude photographs from users and really is putting user safety
first. I have not done a study on this, but anecdotally I know that a lot of
women prefer that app because of how safe they feel there. It is
constantly introducing more features for safety.
Another interesting comparison is this plug-in called Block Party, which is
available on Twitter. Block Party allows users to segment off certain
abusive content, so that you do not have to look at it and deal with it. In
fact, you can even assign someone to deal with it, report things and go
through it for you. It is a very useful app for those who have a lot of
mentions and receive a lot of vitriol. I know a lot of reporters who use it.
Twitter has actually started to incorporate some of the means by which
you can filter and block that content into its own platform, so that is an
instance in which an open platform, such as Twitter, allowing a plug-in
such as that to be developed has led to policy change at the platform
level. Facebook does not allow that. Opening up the API of these
24
platforms in order for user safety applications to be developed would be a
positive development.
Matt Harrison: It is definitely a very interesting idea, and I could see
many benefits, as you say. I am not a competition expert, so I cannot
really delve into too many of the details about it. As a broader point,
there is a commercial aspect already in existence for social media
platforms. It is a conversation we have across all products and services,
in the sense that learning disabilities have a community of 1.5 million
people. Then you start adding on friends, family and acquaintances and
suddenly you have quite a large market.
I am still quite scratching my head about why one of the social media
platforms has not made it a mission to do everything it can to make it the
safest space for someone with a learning disability to be in. You can see
the business benefits from that model. It is not necessarily a piece for
this legislation, but there is also a business case for actually acting on
these online harms and making them safe spaces. I am still quite
surprised that none of them seems to have realised that.
Clare Pelham: To supplement that point, I suspect that we will get
greater competition following better regulation. When the cost of entry
and policing becomes higher, the economic consequence that you would
expect is broader consumer choice. All of us who are parents, as well as
being concerned about our friends and neighbours, would welcome the
advent of a social media platform that we understood to be safe. That
would be a wonderful outcome from the Bill.
Lord Gilbert of Panteg: It could be the same social media platforms,
but just used in a way that is safe for you, so you create your own totally
bespoke environments in the same place as your friends, your family and
everybody else you want to talk to. It would be far more useful for a
social media platform to sit down with you and create an app for your
members and the people you advocate for to use the internet totally
safely, given their condition, than to give you £75,000 to clear it up.
Clare Pelham: If only it had been £75,000. It was £7,500, to be clear.
Lord Gilbert of Panteg: There we are. I gave them much more credit
than I normally do.
Q62 Dean Russell MP: Izzy, we have heard some really difficult testimony
today, especially on harms for children, and I want to build on a question
from earlier. Why do you think the big tech platforms do not seem to care
about safeguarding children in the way that they should?
Izzy Wick: There is an element that ignorance is bliss. The minimum age
of use for most social media platforms is 13, but we know that a huge
number of under-13s use these platforms. If the companies recognised
this, their use base would drop quite significantly. As Clare said, it is
about following the money. It will not have escaped the committee’s
notice that the age-appropriate design code came into force earlier this
month and with that regulation has come a huge raft of changes, such as
25
removing auto-play on YouTube for Kids and minimising the amount of
data that is collected on children. So regulation works and, unfortunately,
it is the only thing that works. These companies have not been
forthcoming in making changes of their own volition, but they absolutely
do when it is required.
Dean Russell MP: Building on that, in terms of the wilfulness of not
caring or seemingly not caring, do you think that we will put this
legislation in place and they will spend billions of dollars and pounds just
trying to find loopholes, so they can continue as they are? What do we
need to do in this Bill so that we are not just forcing them and bringing
the horse to water to drink, as it were, but so that they actually want to
make sure that safeguarding is there for our kids?
Izzy Wick: I sincerely hope they do not do that, but the key to this is
compliance and enforcement. Without wanting to sound like a broken
record, this comes back to minimum standards. This is something that
not just users but actually the companies themselves are really crying out
for.
I will give the example of age assurance. At the moment, everyone has
got into a big muddle about age assurance, and people are rightly
concerned about privacy implications, as are we when it comes to
children’s privacy, but age assurance is not the same thing as
identification, and you can establish a user’s age without knowing
anything else about them.
The technology exists. What is missing is the governance around it.
Industry, as well as parents and children, is saying, “We want rules of the
road to understand what the requirements are, when it’s needed and the
standards that it needs to meet”. Until we have that governance and that
framework, there will not be trust in the system.
Clare Pelham: To answer your question directly, the single most
impactful thing you could do would be to make the sanctions on social
media senior managers include, in the ultimate, following repeated
breaches, a custodial sentence. To paraphrase what someone once said,
once you have them with a custodial sentence, their hearts and minds
will follow.
Dean Russell MP: Nina, talking about politics and the impact especially
on women in politics, a phrase was heard earlier. Matt called it
“dogpiling”. I call it hate mobbing, where an individual will say something
on social media, and all of a sudden a hate mob will appear out of
seemingly nowhere and attack that person en masse, and then move on
to the next target. I have seen it happen to my female colleagues from
across the Floor in particular, but to all politicians.
From your experience, especially based on testimony we had the other
day, which indicated that the same groups of people seemed to be doing
a lot of the targeting on these things, do you think there is a concerted
effort in the space of politics to try to take down certain individuals, or is
26
it just that society happens to get together and criticise these people en
masse?
Nina Jankowicz: It is absolutely a concerted effort. In our research, we
have seen that one individual will post a piece of content, whether on
Twitter or on Facebook, with a large following, and in that original piece
of content there is nothing that violates terms of service, but that is the
dog whistle that initiatives the dogpile and then all the vitrioling content
comes from there.
We have had network analysis done by my colleague Alexa Pavliuc, who
has visualised how this content moves around on these networks. The
networks can see this. If they wanted to, they could visualise this
themselves, see where it is coming from and enact penalties because of
that. That initial piece of content did not trip the tripwire, but these folks
who are initiating the dogpile we believe should and can be punished,
because they are unleashing what is much worse. It is the quantity,
again, of that content coming to you and it is concerted.
It follows different patterns and networks. They are constantly being
reported over and over, and nothing is happening. That is what is most
disheartening about trying to report. I know a lot of folks who do not
even report any more because they do not see the point. They do not
want to waste their time. They do not want to retraumatise themselves.
It is within the capacity of the social media platforms to see these
networks, see how it is moving and see who the repeated offenders are,
and they need to be compelled to do so.
Dean Russell MP: Thank you. Within the context of this Bill, often when
we talk about these large hate mobs, as I call them, we are thinking
about people, but do you think from your analysis that it is not just
people but it is actually triggering a lot of bots that are not real people,
and helping to put lots of content online that is written by AI bots, which
are helping spread that misinformation or that hatred.
Nina Jankowicz: In our analysis, which again was a moment in time of
just two months and quite a lot of content, we did not see that much
computational propaganda—the artificial creation of bot accounts or
things like that. We did see some instances of repetitive posting—certain
individuals posting over and over on a Reddit subreddit, tweeting at a
certain individual over and over or potentially using sock puppet accounts
to do so.
I have seen that in my own experience, where someone will come at me,
log in to their other account and say something else, and then log in to
another account. It is a remarkable amount of effort and, again, that is
what is so disheartening, because these are real people behind it. If it
was bots, you would be able to say very easily, “Okay, here are 25 eggs
that have just tweeted the same curse word at me”, but these are often
real people.
27
To come back to the anonymity question, which we mentioned before, I
have actually had people very happily abusing me under their legal
names, in fact even on LinkedIn, where their employers might see it. I do
not think that introducing a real identity requirement would stem at least
the misogynistic abuse that we see. It is so endemic that people are quite
happy to attach their name to it and, as you have heard in previous
sessions, it would endanger activists in many countries, so, for me,
addressing anonymity does not solve this issue at all.
Dean Russell MP: If you had all the heads of all the big platforms here
and you were to say that there was one bit of the Bill that you absolutely
want to see get through, or a bit of the Bill that is missing that you would
want to see, to make sure you are holding them to account, what would
it be?
Matt Harrison: I spoke before about explicitly listing the protected
characteristics in the Bill and, as I talked about before, putting that
towards the top in the priority content list. Then that flows throughout
the Bill and it gives a sense of direction to everyone—to the Government,
to Ofcom and to the platforms—that when you look at the codes of
practice, sitting in the background are already those protected
characteristics explicitly. When you start looking at phrases like content
that is harmful to adults, you already have a sense of direction about the
characteristics that need to be explicitly protected. It takes out all the
ambiguity and then you can actually start to work with Ofcom and the
social media companies from a slightly stronger point of view. That is the
one big ask that we would have.
Nina Jankowicz: I talked about this a lot, so I will keep it brief.
Transparency in reports, take-downs and other decisions is incredibly
important in order to understand how large the problem is and what, if
anything, the social media platforms are doing to stop it.
Izzy Wick: I have 14 things that I want to change about the Bill, but I
will stick to one.
Dean Russell MP: Just give one for now.
Izzy Wick: Children have a right to protection wherever they are online,
but in the current draft of the Bill we have a definition of regulated
services that includes only user to user or search services. This will leave
a huge number of services that children access on a daily basis out of
scope of the Bill. I am talking about app stores that routinely misadvertise the minimum age of the use of apps and commercial
pornography sites that might not host user-generated content. The status
of edtech is also unclear. Without bringing those things into scope, there
are just huge corners of the digital world that will not need to comply
with the Bill’s safety objectives. There is a very simple solution to this,
which is to include in the definition of a regulated service any service that
is likely to be accessed by children.
28
Ian Russell: I am going to pick the sharing of data for bona fide
research. We are all slightly working in the dark, because it is very
difficult to know what research has been done on this and for what
reasons, and it is about time that we did have a clearer picture of the
effects of the online world on the people who use it. It is very hard to
gain that at the moment, because some of the research is paid for by the
tech platforms and, whether or not there is a conflict of interest, there is
a possibility of that existing and it is very hard to know how you can
judge that.
I would like to see the tech companies, which are effectively data-mining
companies—let us face it—being compelled to give anonymised data, so
there are no privacy issues, to bona fide researchers, so that up-to-date
and constant research can be done and we all know where we are on this
issue.
Clare Pelham: I just have one ask for the committee, really. Please give
us Zach’s law, as recommended by the Law Commission. Please make it
a criminal offence to deliberately send flashing images with the intention
of causing a seizure. If you do, 600,000 people with epilepsy will be
grateful to you.
Q63 Lord Black of Brentwood: I would just like to pick up on the point you
made, Mr Russell. It also comes down to something you said earlier,
Nina. You go to Facebook and it says, “Well, 50% of X or 75% of Y”, but
you never really know what is going on. It would be good to hear from
you about your existing relationships and efforts to get data out of the
platforms, data being so important in this area. We heard powerful
evidence last week from the FA, Stonewall and the Antisemitism Policy
Trust about how they try, but they cannot get it. The platforms will say
nice things and there will be warm words, but then nothing ever happens.
Nina Jankowicz: I will start with that, piggybacking off the report that
we did. You might have realised, if you took a look at it, that Facebook is
not one of the six platforms that we analysed, and that is because of the
lack of access to data. We looked at three alternative platforms and three
more mainstream platforms. The ones we chose were the ones that had
open access, because otherwise you are comparing apples and oranges
again.
Facebook has its CrowdTangle social listening software, which may be
under threat of being shut down. It is how most researchers understand
what is going on on the platform, but in that we only have access to 2%
of the total content on Facebook, because it is only public content that is
indexed there. We do not know the degree of abuse that was perpetrated
in the two months ahead of the US 2020 election in private or in private
groups. It is staggering when you are a member of these groups—many
journalists and researchers will create their own sock puppet accounts to
listen and look there—to see what is not being reported on.
Facebook will say, “It’s just this group” or “It’s just a couple of posts”,
but we really have no idea and it is very difficult to gauge. That is why
29
not only is access to researchers for that data extremely important, as
long as it is pseudonymised, but those transparency reports will provide a
lot of detail and richness that we frankly do not have, and will allow us to
hold social media platforms to account rather than them banner waving
every time they do a take-down that they are proud of.
Ian Russell: It is difficult for me to comment, because a lot of the data I
have seen in most detail is part of the ongoing inquest process, so I
would just say that, as that process is ongoing, I hope the committee will
pay attention to it as the coroner reports. A lot of really important
information will come out of that process, because the volume of data
that the coroner has called into his inquest is unprecedented. We hope to
learn a lot there, so that is hopeful.
Outside of that, I have just one piece of evidence about one incident, the
tiny part that I played in helping Alice Hendy. I do not know how many of
you may know Alice Hendy. She started Ripple tools, which launched a
Google Chrome browser plug-in, so that if people were searching online
for suicide or self-harm terms, before they were connected to the result
of that search, a more prominent banner would come up offering services
for help and support, which is an amazing piece of technology. She has
pioneered this within a year, sadly, of her brother Josh’s death by suicide.
I was present at one meeting where Alice was working with a prominent
search engine and trying to persuade it that it would be good for it to
integrate this. It was surprising to see the resistance of that search
engine to this technology. It seems fairly obvious that it would do good
and help to prevent harm and suicide. It was surprising to hear that they
were asking me for evidence for hashtag search terms, for example, that
people might use. I do not know why they would be asking me that. They
should know that, because they have all the data.
Matt Harrison: It is probably quite difficult. I cannot think of a time
when we proactively asked for a large dataset to go through from social
media companies, but I know that it has not been proactively offered to
us. Most of our dealings have been more about the generic policies, the
top-level policy lines or exploring ideas for promoting good content
online. We would be fascinated to see that data and to get a better
understanding.
There is something to be said about the difficulty in understanding data
on disability, because some people might not identify in their account as
disabled and might be afraid to post content that might flag them as
someone with a disability. So there are slight issues around the data
there, but we would be really intrigued to see what they do have.
Clare Pelham: The piece of data that I would really love to have is how
many are among the silenced—the people who have closed their social
media accounts, who have been driven off these platforms because of the
cruel and hateful messages they have received, who have been excluded
from our national conversation and marginalised. I know that many of
those who have closed their accounts have done so with messages to