The Risk of AI Increasing Gender Inequality: What Opportunities Exist to Bridge the Gap?

Without careful supervision of input data, AI can reflect and even amplify existing gender biases in society.

Improper use of AI can widen the gender gap and lead to online gender-based violence

AI has become a familiar assistant in our daily lives, but it carries inherent risks to gender equality if the technology is trained and deployed using data that reflects a biased society. In such cases, AI models lose their neutrality and instead become tools that reinforce and legitimize gender stereotypes, generating content characterized by hate, contempt, and the degradation of women.

According to a UNESCO study, Large Language Models (LLMs) frequently associate women with caregiving, domestic roles, or "behind-the-scenes" work, while men are linked to power and technical professions. This bias is also evident in AI image generators like Stable Diffusion or Midjourney; a study conducted in March 2025 found that men appear overwhelmingly as doctors, engineers, and managers, whereas women are often over-juvenalized or depicted through stereotypes of being "attractive" or "gentle."

This bias becomes dangerous when AI is applied to decisions with direct life impacts. For instance, in recruitment, many AI screening systems undervalue female candidates simply due to career breaks for childbirth or family care, regardless of their actual professional capabilities.

The misuse of AI is also making online gender-based violence increasingly sophisticated. Deepfakes are a concerning example; 90–95% of deepfake content online is non-consensual pornography, and approximately 99% of the victims are women. Furthermore, according to a UN Women report and analysis released in November 2025, AI is being exploited for impersonation, sextortion, and the creation of personalized harassment campaigns targeting women, causing widespread and uncontrollable psychological harm.

Ms. Vu Thi Hanh, an AI Expert at the Viettel Data Service and Artificial Intelligence Center (Viettel AI), remarked: "Gender-based violence does not only occur physically in real life; it is spreading rapidly across cyberspace. AI can become a tool exploited by perpetrators if we do not have the regulations, skills, and solidarity to protect one another."

Ms. Vu Thi Hanh (Viettel AI) shares insights on gender inequality in the digital space

Mastering AI to Safeguard Gender Equality and Build a Safe Digital Environment

To ensure AI becomes a tool for narrowing rather than widening the gender gap, solutions must be implemented simultaneously based on three pillars: developers, the user community, and the social ecosystem.

For tech companies, periodic assessments of model bias must be conducted both before and during the deployment of AI systems. Accordingly, risk assessments should be performed to ensure that training data is diverse in terms of gender, age, nationality, and background, limiting AI models from learning from biased data patterns and delivering unfair results. Platforms must also be held accountable for toxic content generated, disseminated, or profited from through their systems.

In addition, enhancing digital literacy for users is a vital factor. Ms. Vu Thi Hanh emphasizes the “5 No’s - 4 Yes’s” principle: No sending sensitive information; No treating AI-generated content as absolute truth; No using AI to violate the privacy of others; No relying on AI to write important documents without verification; No using AI to create or edit images and videos to defame or insult anyone's honor; and Yes to verifying information; Yes to protecting accounts; Yes to taking responsibility for the content you create; Yes to sharing experiences on using AI safely. These principles help users build a protective fence for themselves against the rapid pace of AI and technological development.

At the societal level, coordination is required between regulatory bodies, tech enterprises, and organizations to ensure that AI is deployed in a safe and gender-equal manner. Training programs on deepfake identification, digital identity management, safe online conduct, and the prevention of online violence need to be expanded across schools, communities, and workplaces. Recently, the National Assembly of Vietnam passed the country's first Artificial Intelligence Act, clearly demonstrating a policy of developing AI that is responsible, transparent, and human-centric.

AI is an inevitable trend. What matters is not stopping the development of this technology, but how humans supervise and harness it. When data is controlled, models are regularly assessed, users are fully equipped with digital skills, and the community joins hands to combat online violence, only then will AI truly become a tool to narrow the gender gap instead of replicating long-standing inequalities.