Queen's University Belfast
Creating a safer academic environment.
Queen's University Belfast
Creating a safer academic environment.
Historic Royal Palaces
How AI moderation protects universities from hate campaigns and scams.
How AI moderation tackles hate speech and spam at scale.
26,032
total comments monitored
2,319
harmful comments moderated
2.3x
increase in engagement
Safeguarding students and staff.
Universities are meant to be spaces for freely debating ideas about how the world should be. So when staff members are targeted through politically motivated misinformation campaigns, that's cause for concern.
Queen's University Belfast came to us with a this problem: their newly appointed chancellor became the target of hate and conspiracy theories. Not just on online forums, but in QUB's social media comments section.
QUB wanted to safeguard their staff members by moderating these comments. This would in turn help preserve their tenor of open and respectful exchange. They simultaneously wanted to optimize the way in which they communicated with their student body, freeing their comments section from spam and student fraud. Their social media pages were intended to be a place to disseminate exciting new research to the world, and connect with their students. Their overstretched social media team needed help to make that happen. The solution: Arwen's customizable AI.
Safeguarding students and staff.
Universities are meant to be spaces for freely debating ideas about how the world should be. So when staff members are targeted through politically motivated misinformation campaigns, that's cause for concern.
Queen's University Belfast came to us with a this problem: their newly appointed chancellor became the target of hate and conspiracy theories. Not just on online forums, but in QUB's social media comments section.
QUB wanted to safeguard their staff members by moderating these comments. This would in turn help preserve their tenor of open and respectful exchange. They simultaneously wanted to optimize the way in which they communicated with their student body, freeing their comments section from spam and student fraud. Their social media pages were intended to be a place to disseminate exciting new research to the world, and connect with their students. Their overstretched social media team needed help to make that happen. The solution: Arwen's customizable AI.
Improving customer experience through safeguarding
Historic Royal Palaces (HRP) is an independent charity that manages iconic landmarks such as the Tower of London and Hampton Court Palace. As a prominent cultural organisation, HRP maintains a strong social media presence to engage a global audience, promote events, and drive visitation. However, as their online presence grew, they faced increasing challenges in managing comment volumes across both paid and organic posts.
Paid Facebook ads posed a particularly severe challenge, as images of diverse families attracted harmful comments, including hate speech and divisive political opinions. With engagement spiking during holidays and weekends, HRP’s small social media team struggled to manually moderate thousands of incoming comments, often falling behind and risking reputational damage. Existing tools were insufficient in accurately detecting harmful content, and the leftover toxic comments prevented their community from being a positive and inclusive space.
HRP recognised the need for an efficient, scalable moderation solution to protect their community. To address these challenges, HRP partnered with Arwen to automate comment moderation and safeguard their social media channels.
.png?width=1352&height=1496&name=Group%206448%20(4).png)
Challenge
QUB's social media accounts were receiving a growing amount of targeted hate comments. Many of them were highly specific conspiracy theories, so they were difficult to catch through simple keyword filters. QUB's social team wanted to make sure that they were only removing truly harmful comments, while allowing students to voice negativity and criticism.
Fraud comments were also clogging up their social channels, with bot accounts offering to courses or services to complete students' work for them. QUB wanted these removed, while still allowing for legitimate links to resources to be shared in the comments.
This meant that out-of-the-box moderation solutions were not advanced enough at discerning unwanted comments to the level that QUB required. They needed something more customizable.
Challenge
QUB's social media accounts were receiving a growing amount of targeted hate comments. Many of them were highly specific conspiracy theories, so they were difficult to catch through simple keyword filters. QUB's social team wanted to make sure that they were only removing truly harmful comments, while allowing students to voice negativity and criticism.
Fraud comments were also clogging up their social channels, with bot accounts offering to courses or services to complete students' work for them. QUB wanted these removed, while still allowing for legitimate links to resources to be shared in the comments.
This meant that out-of-the-box moderation solutions were not advanced enough at discerning unwanted comments to the level that QUB required. They needed something more customizable.
.png?width=1352&height=1496&name=Group%206448%20(4).png)
Challenge
HRP's small social media team of three faced significant challenges in moderating an overwhelming amount of comments due to their growing social media presence, particularly during weekends and holidays. Paid Facebook ads proved especially problematic, attracting up to 30 harmful comments per ad, including hate speech and divisive political content that could damage HRP’s reputation and discouraging audiences from engaging with ad campaigns or visiting their sites.
Manual moderation and inadequate tools made it difficult to differentiate between harmful and appropriate engagement. HRP’s workflow left little room for proactive audience interaction, as valuable time was spent removing spam and toxic content. Senior leadership grew concerned about the reputational risks posed by visible hate speech and spam on ads, particularly campaigns featuring diverse families or heritage-related topics.
Resource constraints further exacerbated these issues. The social media team managed 13 profiles across multiple platforms, and seasonal peaks during holidays created additional pressure. Weekend coverage relied on PR teams, who lacked the bandwidth for detailed moderation, leaving HRP struggling to maintain a positive and welcoming online presence.
.png?width=1807&height=2000&name=Group%206448%20(2).png)
Solution
- Arwen used a combination of highly trained models and advanced pattern detection to identify conspiracy theories, threatening, scamming, and hate comments.
- These models could distinguish between simple criticism or negativity, which were allowed, and targeted hate campaigns, which weren't.
- QUB implemented these moderation settings across Facebook, Instagram, TikTok, and X.
- Unwanted comments are now identified and immediately removed, leading to a safer, more reputable social media presence across networks.
.png?width=1314&height=1496&name=Group%206449%20(3).png)
“Arwen's changed the game for us. I can't believe we didn't use it before."

.png?width=1954&height=1642&name=Group%206271%20(8).png)
Impact
Arwen's solution means reduced harm for QUB's staff and students. It has also transformed QUB's social media pages into more constructive online spaces, both for students and the global audience interested in their research. This in turn has had a positive impact on their engagement.
- Since QUB switched Arwen on, it has monitored over 26,000 comments, moderating 2,319 toxic and spam comments in real time.
- Auto-moderation has freed up QUB's team to focus on generating engagement.
- Arwen's solution has also created a more welcoming online space, allowing students to interact with their posts more.
- All of these factors have led to over double the amount of positive engagement on QUB's posts, transforming their pages back into a useful hub for interacting with their community.
Impact
Arwen's solution means reduced harm for QUB's staff and students. It has also transformed QUB's social media pages into more constructive online spaces, both for students and the global audience interested in their research. This in turn has had a positive impact on their engagement.
- Since QUB switched Arwen on, it has monitored over 26,000 comments, moderating 2,319 toxic and spam comments in real time.
- Auto-moderation has freed up QUB's team to focus on generating engagement.
- Arwen's solution has also created a more welcoming online space, allowing students to interact with their posts more.
- All of these factors have led to over double the amount of positive engagement on QUB's posts, transforming their pages back into a useful hub for interacting with their community.
.png?width=1954&height=1642&name=Group%206271%20(8).png)
Impact
Arwen created a safer and more engaging space for HRP’s global audience, enhancing user experience and protecting their brand reputation. By automating moderation, HRP could confidently run campaigns without fear of harmful content discouraging potential visitors or derailing conversations.
- Since implementation, Arwen has monitored over 36,000 comments, moderating 2,660 toxic and spam comments effectively.
- With consistent moderation across 13 social media profiles, including Facebook, Instagram, TikTok, and X, HRP now has reliable, real-time protection against toxicity and spam.
- This improved moderation has enabled HRP to enhance their audience’s experience, while their team operates with greater efficiency during peak periods, such as holidays and high-engagement campaigns.
.png?width=2000&height=1672&name=Group%206271%20(7).png)
Queen's University Belfast has taken the next steps.
Are you ready to take yours?