Algorithms function as gatekeepers of the internet, shaping search results, social media feeds, and even financial decisions. However, these systems are not inherently neutral. They are trained on vast datasets that reflect historical and societal biases.
Gender discrimination and the harassment of women is now exacerbated by a new, innovative tool: biased algorithms in the online spaces.
The United Nations representative in Sri Lanka Marc André Franche said this at a recent meet.
“Newer threats, such as biased algorithms and programming inequalities, into online spaces are opening new arenas for harassment and abuse,” Marc André Franche said.
Algorithms can be trained to reflect past narratives and social agendas, she said, explaining, “At the current rate, it would take 134 years to achieve full gender equality.”
A good instance, the UN official said, was how algorithms and artificial intelligence tools have been reported to be downgrading women, steering them into lower wage positions while targeting better job prospects for men.
Online harassment against women is not only pervasive but is also exacerbated by biased algorithms, she said. In Sri Lanka and beyond, these algorithmic systems — designed to optimise engagement and personalise user experience — often reinforce harmful societal norms, making digital spaces hostile for women, she explained.
Recent reports from the United Nations and other advocacy groups shed light on how these biases manifest and what can be done to mitigate them.
The Invisible Bias in Algorithms
Algorithms function as gatekeepers of the internet, shaping search results, social media feeds, and even financial decisions. However, these systems are not inherently neutral. They are trained on vast datasets that reflect historical and societal biases. When these biases go unchecked, they become ingrained in the technology itself, reinforcing discrimination and amplifying online harassment against women.
Marc André Franche said that these AI systems, trained on past data, often perpetuate existing power structures, further marginalising women in online spaces. The effects of these biases are particularly visible on social media, where content moderation policies often fail to keep up with the sheer volume of harmful content being amplified.
Social media platforms, which rely on complex algorithms to drive user engagement, are particularly culpable in the spread of online harassment. Several studies have demonstrated that misogynistic content — ranging from gender-based hate speech to deepfake pornography — is disproportionately amplified by recommendation engines. A report published by The Guardian found that videos containing misogynistic or defamatory content about women were significantly more likely to be recommended compared to neutral or positive content.
Echo Chamber of Hate
One of the most concerning aspects of algorithm-driven social media platforms is their ability to create echo chambers that reinforce and escalate harassment. When a user engages with misogynistic content, algorithms detect this behaviour and begin promoting similar material, creating a self-reinforcing cycle that normalises online abuse against women.
For example, women who are active in public discourse — journalists, activists, and politicians — often find themselves targeted by coordinated harassment campaigns. The more engagement such content receives, the more likely it is to be promoted by platform algorithms, increasing the reach and intensity of the attacks. This pattern discourages women from speaking out, effectively silencing them and limiting their participation in online discussions.
The ways in which these biases manifest are varied and insidious. Algorithms allow harassers to target women based on their online presence, interests, or even physical appearance. Social media companies, in their pursuit of maximising user engagement, have inadvertently created an ecosystem where abusive behaviour is rewarded with visibility. Additionally, misinformation about women — ranging from false accusations to harmful stereotypes — spreads rapidly, bolstered by algorithms that prioritise sensational and emotionally charged content.
Another issue is the failure of content moderation systems to recognise and remove misogynistic content. Many platforms employ automated moderation tools that are often ineffective in detecting context-specific abuse. This results in an environment where harmful content flourishes, while women’s responses or attempts at self-defence are disproportionately flagged and removed.
Real-World Consequences of Digital Harassment
The effects of online harassment are not confined to the digital realm. Women subjected to cyberbullying, doxing, or revenge pornography often experience severe psychological distress, including anxiety, depression, and PTSD. Many are forced to withdraw from online platforms altogether, limiting their personal and professional opportunities.
A Sri Lankan activist who faced relentless online abuse shared her experience: “It’s not just words on a screen. It’s about the feeling of being constantly watched, judged, and threatened. It affects every aspect of your life.”
There have also been documented cases where online harassment has led to physical violence. In some instances, women who were targeted online have been stalked or assaulted offline, demonstrating the tangible dangers posed by algorithmic bias. Women in Sri Lanka who have been vocal about political or social issues have reported being harassed both online and in person, with threats escalating to real-world violence.
Beyond the psychological toll, the economic impact is also significant. Women who rely on online platforms for their work — whether as influencers, journalists, or entrepreneurs — often find themselves disproportionately affected by harassment. When they are forced to reduce their online presence for safety reasons, it directly impacts their professional growth and financial stability.
Algorithmic Discrimination Beyond Social Media
The issue extends beyond social media platforms to other areas, including financial services and job recruitment. In many cases, algorithms used for credit scoring or hiring decisions reinforce gender biases. For instance, AI-driven credit assessments have historically undervalued women’s financial reliability, limiting their access to loans and economic independence. Similarly, recruitment algorithms trained on male-dominated data often prioritise male candidates, perpetuating workplace gender disparities.
Women entrepreneurs in Sri Lanka, for example, have reported difficulties in securing business loans due to algorithmic biases that favour traditional male business models. These biases further entrench economic disparities, making it harder for women to achieve financial independence.
Tackling this issue requires a multi-faceted approach that involves governments, technology companies, and civil society. A crucial step in addressing algorithmic bias is ensuring that the data used to train AI systems is diverse and representative. Tech companies must acknowledge that their algorithms are not neutral and take active steps to identify and rectify biases.
Transparency is also key. Tech companies should be required to conduct bias audits and disclose how their algorithms function. This would allow researchers and policymakers to better understand the mechanisms driving online harassment and work towards solutions. Without transparency, holding these companies accountable remains difficult.
Moreover, stronger content moderation policies are necessary to curb online harassment. This includes hiring diverse teams of human moderators who can assess content in a nuanced way, rather than relying solely on flawed AI-driven moderation systems. Social media companies must also improve their reporting systems to ensure that victims of online abuse receive timely and effective responses.
Gender Equality in the Digital Age
Education and awareness are equally vital. Digital literacy programs can help users better understand the biases embedded in online platforms and empower them to navigate digital spaces more safely. Schools, workplaces, and community organisations should incorporate discussions on digital ethics and responsible AI use into their curricula.
Legal frameworks must also be strengthened. Governments need to implement and enforce laws that criminalise online harassment and hold perpetrators accountable. Sri Lanka, for instance, has made strides in enacting legislation against cyber harassment, but enforcement remains a challenge. A more coordinated effort between law enforcement agencies, tech companies, and civil society organisations is needed to ensure that perpetrators are held responsible for their actions.
Recognising the intersectional nature of online harassment is also crucial. Women of colour, LGBTQ+ women, and women with disabilities often face compounded forms of discrimination that require targeted solutions. Policies aimed at addressing algorithmic bias must take these unique experiences into account to be truly effective.
The fight against online harassment and algorithmic bias is ultimately a fight for gender equality in the digital age. As Sri Lanka and other nations grapple with the realities of a rapidly evolving digital landscape, ensuring that technology serves all users equitably must be a priority.
By addressing the biases embedded in algorithms and fostering safer online environments, we can transform the internet into a space where all women can participate freely, without fear of discrimination or harassment. The responsibility lies not just with policymakers and tech companies, but with society as a whole, to demand a digital world that upholds fairness and equality.
