AI Monitoring: Protecting Professional Players

FIFA Threat Matrix Report

Purpose of the Report & Main findings

This study examined over 406,987 social posts across Twitter and Instagram targeting players and coaches for the EURO 2020 Final (England v Italy) and AFCON 2022 Final (Senegal v Egypt). The study period ran from the end of the semi finals until 3 days after each final.

FIFA Anti Discrimination Infographics

Main findings

VolumeOver 55% of players in both EURO 2020 and AFCON 2022 Final’s received some form of discriminatory abuse.
TypeHomophobic slurs most common form of detected abuse, with racism second.
RaceBukayo Saka and Marcus Rashford (England) were the most abused players in the EURO 2020 Final / Mahmoud Hamdy (Egypt) was the most abused player at AFCON 2022 Final’s.
PlatformsAbuse on Twitter is constant across the period whilst Instagram abuse is more event driven – i.e. losing final. Tactically, over 75% of Instagram comments and abuse included emojis.
GeographyMajority of abusers come from the players home nation.
Club affiliationPlayer club identity is a trigger for abuse. i.e. Mohamed Salah received abuse on Twitter from fans who are supporters of Liverpool’s English Premier League rivals.
Officials/ManagersAFCON coaches received double abuse of EURO 2020 managers.

Report Details
The EURO 2020 Final accounted for 365 abusive posts, whilst the AFCON Final accounted for 149.
Types of Abuse

The study revealed 514 abusive posts targeting players. Racism and homophobia were the two most prevalent forms of abuse present across the tournaments representing a combined 78% of all detected abuse on Twitter and Instagram.

Targeted Players

This study focused on the number of abusive accounts that could be identified – showing that more action can be taken by authorities, national football associations and (where there is a clear affiliation) domestic clubs.

Areas of Focus & Recommendations

HOMOPHOBIA This study further highlights the growing problem of homophobia surrounding football, and a failure to tackle the issue on social media. A heavier focus on campaigns around this issue ahead of the upcoming World Cup would be recommended - along with more activity from the platforms in detection of this category of online discrimination. NUANCE The nuance of issues for global and regional tournaments, such as racist abuse highlighted in this study, means universal moderation solutions from platforms are unlikely to help tackle issues in a meaningful way. It must be applied with the knowledge base and contextual understanding of parties like FIFA who know the sport, players and fans as well as the issues. DOMESTIC > INTERNATIONAL Understanding the link between domestic clubs and abuse at international competitions is warranted by the output of this study. Better information supporting this issue can be filtered down from FIFA to national associations. PROACTIVE MONITORING Procative monitoring can identify, categorise and detect abuse on social media. Owning this process empowers FIFA to take more action. Equally FIFA’s strategy should seek to take knowledge from domestic studies to flag issues likely to impact tournaments. ACCOUNTABILITY Holding platforms to account is still essential, identifying the volume and type of abuse on each platform can support their efforts and ensure football is playing its part in tackling this problem. It is also important to audit their action in tackling abusive or threatening content. DATA BENCHMARKS Data benchmarks provide the ability to understand whether the situation is improving or getting worse.


Threat Matrix provides real-time monitoring and analysis of millions of open source social media posts across multiple platforms.

Source DataEvidence based reports and recommendations covering high-level network reporting with detailed verification. Cross-referencing digital + real world intelligence.
Process DataClean all data (removing bots) before applying AI Natural Language Understanding. Filtering of abuse or threat to identify problematic messages.
Human ReviewDeliver a meaningsbased assessment based on Augmented Intelligence. The speed of AI + the nuance of human interpretation.
Actionable OutputEvidence based reports and recommendations covering high-level network reporting with detailed verification. Cross-referencing digital + real world intelligence.