Gendered disinformation and the path from online narratives to offline harm

16 November 2026
Gendered disinformation and the path from online narratives to offline harm

Technology-facilitated gender-based violence (TFGBV) – including cyber harassment, image-based abuse, and gendered disinformation, misinformation and hate speech – has increased significantly over the past decade. Among the main targets of TFGBV are women politicians, activists, human rights defenders, refugees and women in conflict situations.

In several contexts, gendered disinformation, misinformation and hate speech are often used as a strategy for:

  • Silencing women or sexual and gender minorities
  • Mobilizing political or military constituencies
  • Undermining social cohesion
  • Fuelling polarization
  • Inciting physical violence, including gender-based violence

Research in Ethiopia and Myanmar, for instance, has shown that social media platforms and algorithms have amplified and accelerated the spread of harmful rhetoric and hate speech. This has contributed to the incitement of violence, including systematic rape and other forms of sexual violence.

In non-conflict settings like Brazil, Hungary, Italy, India and Tunisia, there have been examples of women politicians becoming targets of gendered disinformation, misinformation and hate speech. The narratives deployed often depict women as foreign agents, unpatriotic or corrupt, resulting in online violence as well as threats of physical violence. 

Coordinated disinformation operations observed in Russia, Serbia and Türkiye, meanwhile, have exploited grievances among marginalized men, linking national decline with gender and sexual equality and presenting it as a foreign ideological colonization in collision with national fundamental values, often for political gain. 

By exploiting harmful gender stereotypes and social norms, gendered disinformation seeks to discredit women and marginalized groups, undermine their credibility, and hamper their participation in public life. This hinders women’s meaningful participation in peacebuilding, mediation, disarmament and conflict resolution efforts, thereby affecting international security more broadly.

In view of all of this, TFGBV, including gendered disinformation, misinformation and hate speech, is recognized as a growing threat for democracy, peace and security.

Working together to overcome challenges

That’s why, in September, EPFL’s EssentialTech Centre convened a hackathon to find solutions to address the problems of disinformation, misinformation and hate speech. This two-day event brought together students and experts from different backgrounds to work on nine challenges proposed by international organizations, NGOs and public entities in Switzerland.

Poster for the EPFL PeaceTech Hackathon 2026 on 26–27 September at the EPFL campus. It shows a woman’s portrait over the repeated words “misinformation”, “disinformation” and “hate speech”, with a digital dove.

During the hackathon, participants were split into teams and had to work together to address specific problems. The groups engaged in friendly “coopetition”, alternating moments of cooperation and competition to come up with creative, practical and feasible solutions, including two that were particularly relevant to this issue. Two of the challenges and the respective solutions proposed by some of the hackathon participants are presented below.

Harm Lens

Online gender-based violence can often escalate into physical violence, including violence facilitated by weapons.

For example, nearly 60% of human rights defenders, activists and journalists have reported experiencing offline attacks connected to digital abuse.  

With this in mind, UNIDIR’s Integrated Approaches Programme challenged hackathon participants to consider how we can detect escalation from online gendered hate to physical harm.

The group responded by developing Harm Lens, a prototype platform designed to monitor misogynistic language and other signals in publicly available online content.

This platform would assign risk levels based on a harm taxonomy, offering an early-warning system for violence against women and LGBTQI+ communities. The platform would also allow users to visualize networked activity between social media accounts, such as cross-posting and other connections within and between particular social media platforms.

Given the current data gap on escalation from online to offline violence, the prototype relied on indicators, scales and ranking systems for misogyny already developed by the Institute for Strategic Dialogue and the Polarization and Extremism Research and Innovation Lab.

Designed to be open source and compliant with EU regulation on data protection, Harm Lens could be used in a wide range of real-world contexts. Organizations working on women’s rights, LGBTQI+ rights, digital safety, conflict prevention and human rights, for example, could use it to monitor emerging patterns of online harassment and identify communities, narratives, or topics that warrant closer attention.

HateWatch Syria

Another interesting solution emerged from a separate challenge proposed by Humanity and Inclusion: “How can we build community-based early warning systems to prevent digital harm?”

The team working on this challenge chose to focus on Syria. After the fall of the Assad regime, the country saw a considerable spike in online hate speech and disinformation campaigns against women and members of minority communities, with threats sometimes escalating into offline violence.

The team responding to this challenge came up with a prototype called HateWatch, a mobile application to be used by Humanity and Inclusion staff in Syria. The app would enable users to securely report violent events in real time, both in English and Arabic. This would include hateful online content (screenshots, photos, videos, documents), coded discriminatory terms, and links to Facebook posts and groups spreading gendered disinformation and hate speech.

Reports would be reviewed by managers to verify their accuracy and analyse incidents. A visual map would help identify hotspots and priority interventions where local mitigation through in-person community engagement might be required to reduce the risk of escalation into offline harm. 

Hackathon participants with laptops discuss around a table during the EPFL PeaceTech Hackathon 2026, beneath a large screen showing the EPFL logo.

What monitoring tools can and cannot do

The speed, scale and reach of harmful content online have made it increasingly important for governments, civil society organizations and international actors to analyse harmful gendered narratives and abuse in real time.

Tools like Harm Lens and HateWatch Syria could help by identifying the targets, volume, frequency and nature of the material circulating online. Indeed, several existing platforms use similar approaches, including Build Up’s Phoenix Platform and Meedan’s Suwali tool.

Often referred to as “social media listening” tools, they use machine learning and large language models to collect, analyse and classify social media content, including posts containing hostile, toxic, hateful, misogynistic or threatening language.

UNDP’s eMonitor+ platform, meanwhile, combines AI-assisted classification with human verification. Already deployed in 39 countries, it includes a dedicated module for monitoring violence against women in elections.   

These tools help identify spikes in gendered disinformation and hate speech, including threats to women, gender-diverse people, human rights defenders, journalists and politicians. They can also enable timely responses by civil society, humanitarian organizations and public authorities.

However, they also come with significant limitations:

  • Language and context: Machine learning models may struggle to accurately interpret local languages, dialects, slang, sarcasm, humour or culturally specific references. This can result in false positives, where harmless content is incorrectly flagged, but also false negatives, where harmful content goes undetected. Here, the role of human reviewers familiar with local context and language becomes crucial.
  • Data access: Some social media platforms have started to restrict large-scale data access by introducing fees for researchers and other external users.
  • Evasion: Tools may struggle to track coordinated operations that deliberately adopt tactics designed to avoid detection, such as coded language or migration across multiple platforms.

Combining digital tools with coordinated policy responses

To complement the deployment of digital tools, States and organizations are also looking to develop policy responses to the problems of disinformation, misinformation and hate speech.

The United Kingdom’s RESIST 3, for example, offers a structured approach for government communicators to identify, assess and respond to information threats, including disinformation, misinformation and foreign coordinated influence campaigns.

But even if efforts to address disinformation, misinformation and hate speech are underway, they still tend to lack coordination and standardization.

This highlights the need for stronger international frameworks and deeper collaboration between governments, the private sector and academia to reinforce information integrity and mitigate related harms.

As we learnt at the hackathon, engaging the widest possible range of stakeholders, including youth, will be essential if we are to generate innovative solutions to these evolving challenges.