Other AI tools, such as virtual assistants, have echoed and amplified gender stereotypes. According to UNESCO, AI-powered voice assistants have been shown to respond submissively to abusive commands, perpetuating harmful norms about women’s roles and behaviors.⁶

Following such revelations, companies like IBM and Microsoft committed to diversifying datasets and improving algorithmic accountability. These are essential steps, but public policy must keep pace. If left unchecked, biased AI could hinder women’s access to employment, education, health and justice.

In addition to reinforcing social biases, AI is also reshaping the labor market—often to the detriment of women. Routine and clerical jobs, where women are overrepresented, are among the first to be automated. Meanwhile, roles in AI development, data science, and Science, Technology, Engineering, and Mathematics (STEM) fields are expanding rapidly, but women remain underrepresented. Globally, only 22 percent of AI and data science roles are held by women.⁷ Without affirmative intervention, this technological shift risks widening economic inequalities. To reverse this trend, national and sectoral policies must prioritize digital skills training and STEM education for women and girls, ensuring they are not just consumers of technology but also its creators and leaders.

Gendered Harassment in the Age of AI

Digital spaces hold the potential to empower—but for many women, they have become places of surveillance, harassment and exclusion. AI-generated explicit content disproportionately targets women, with over 96 percent being non-consensual and often used for defamation or coercion.⁸ In Pakistan, manipulated AI imagery is now implicated in 30 percent of reported cyber harassment cases, with women—especially those in the public eye—bearing the brunt.⁹ Alarmingly, women account for nearly 90 percent of all harassment complaints filed with Pakistan’s Federal Investigation Agency, highlighting the urgent need for stronger digital protections and accountability.

The consequences are chilling. About one-third of female students and nearly half of working women report harassment online.¹⁰ Many are forced to consider leaving school or the workforce entirely. The Prevention of Electronic Crimes Act (PECA) 2025 attempts to address these threats, but the law lacks clarity and could be misapplied in ways that stifle speech rather than safeguard victims.¹¹

Towards a Gender-Inclusive AI Future

The 2024 Summit of the Future marked a turning point. The Pact for the Future and its companion framework, the Global Digital Compact, called on governments to remove gender-based barriers to technology and ensure that emerging tools like AI are shaped by—and serve—all people equally.

UNDP contributed to this global conversation from the ground up. In 2024, during national consultations ahead of the Summit, civil society in Pakistan issued the Islamabad Declaration, urging a people-centered approach to digital transformation—one that treats digital access not as a privilege, but as a public good.

Our Doing Digital for Development report confirms this: districts with higher digital adoption see significantly better human development outcomes.¹² That is why UNDP’s ‘4As’ framework—Access, Adopt, Anticipate, and Accelerate—centers equity, especially for women and marginalized communities, as the cornerstone of meaningful digital progress.