Social scientists are helping the Welsh and UK Government tackle the over 140,000 hate crimes recorded by police in England and Wales in 2024, by using AI to detect online harms.
Summary
HateLab is a global data hub using social science and computer science expertise to help identify and tackle hate speech and crime. Nisien.AI is a spin out from this which uses the latest AI algorithms developed within HateLab to accurately detect and classify online harms across platforms in real time and provides users with guidance on how to respond and counter online harms effectively. This research has contributed to the Welsh Government’s national Framework for Action on Tackling Hate Crime and been directly embedded into the UK National Cyber Hate Crime Hub contributing to improved response times, better support for victims, and more effective allocation of resources.
Using AI to tackle online hate crime
“The provision of the HateLab platform, co-created with the Hub, has fundamentally changed the way we monitor the spread of hate speech during national events.”
National Online Hate Crime Hub
The challenge
Between 2012 and 2023 police recorded hate crimes in England and Wales rose by 252% according to Home Office Research. Over 50% of people over 12 are routinely exposed to hateful content online, not only posing risks to mental and physical wellbeing, but hateful narratives also potentially trigger and escalate hate crimes offline and exacerbate community tensions. Tracking, analysing and assessing harmful content online is expensive and work intensive, making it difficult to assess the level of potential threats and respond proactively.
The research
HateLab, based at Cardiff University, is a global hub for data and insight into hate crime and speech. Using a multidisciplinary mix of social science and computer science, the research team draws on a range of methods including surveys and machine learning, and criminological concepts to generate a broad evidence base on hate crime both on and offline in Wales, the UK and beyond.
The foundations of HateLab came from the All Wales Hate Crime Project, which remains the largest and most comprehensive study of hate crime in the UK, surveying over 1,800 people and interviewing over 60 victims, producing a robust source of data which could be used by policymakers in Wales for the first time.
Following this, the HateLab dashboard, which uses machine learning to detect online hate speech in real time and at scale, was developed to assist researchers, practitioners and policymakers to monitor the spread of online hate speech. It is also used to provide counter-speech to repeat perpetrators. The dashboard is provided free of charge to civil society organisations that do not have the resources to routinely monitor and counter online hate speech and divisive disinformation themselves.
The Director of HateLab, Professor Matthew Williams, alongside collaborator Professor Pete Burnap, founded HateLab’s spin out Nisien.ai which uses the foundations of HateLab research to develop further tools that organisations can use to anticipate, identify and respond to online threats. As part of this they developed HERO, the Harms Evaluation & Response Observatory, which uses AI algorithms to identify, respond to and reduce a wide array of online harms.
Hear more from Professor Matthew Williams in this episode of the We Society podcast.
The impact
The insights from the All Wales Hate Crime Project directly informed the Welsh Government’s national Framework for Action on Tackling Hate Crime – the first of its kind in Wales. The Welsh Government cited the research as being fundamental in providing insights on evidence on the impacts of hate crime on victims living in Wales across protected characteristics; victims’ experiences of the criminal justice system; and the emerging trend in online hate speech. The findings from the project have also been used to create online hate speech guides for use in schools, campaign materials for National Hate Crime Awareness Weeks, and a Tackling Hate Crime Toolkit, which was distributed to all Housing Associations in Wales.
The HateLab dashboard, which is designed to facilitate near-real time responses to online hate, has been integrated into the UK National Cyber Hate Crime Hub which is run by the National Police Chiefs’ Council and produces intelligence reports for police, senior civil servants and MPs. By providing the police with data on the spread of online hate, the dashboard has directly contributed to improved response times, better support for victims, and more effective allocation of resources.
The HERO suite by Nisien.ai has been used by numerous organisations to tackle online harms including a large social media platform which utilised the team’s expertise to inform their community guidelines to develop new strategies to mitigate online hate speech; and a UK Government department used HERO to monitor anti-migrant content related to the settlement of Ukrainian refugees and provided data that fed into national threat assessments on related extremist activity.
“For us, the value of the platform was not just in identifying hate speech, but also to monitor where tensions did not result in hate speech online. An example of this would be the migration of Ukrainians to the UK following the Russian invasion in February 2022. We used HateLab to monitor signs of any negative reaction to the news that Ukrainian families would be relocated to Wales, as there was a chance that far-right groups might capitalise on the situation with anti-migrant rhetoric. Concerns were increased following an incident in Hermon, Pembrokeshire where Ukrainian flags were torn down, as reported by the BBC. While we ran these queries, HateLab did not pick up any examples of hate speech online via the queries we set up. This tallied with the general feeling of goodwill towards Ukrainians that was supported by numerous other sources.”
Welsh Government