The relationship between technology and society has never been one of passive adoption. Every technological object carries assumptions about who will use it, under what conditions, and with what risks. These assumptions are often treated as neutral because STEM (Science, Technology, Engineering, Mathematics) itself is framed as objective and apolitical. In practice, however, technologies designed by a homogeneous class – often white, male, and privileged – codify social hierarchies into the very architecture of our digital world, treating the male experience as the “default” while rendering the specific risks faced by women and other marginalized groups invisible.
The Male “Default” in Digital Design
This exclusion is reflected in a persistent demographic stagnation within the industry. Despite growing public attention, over the last decade, women have held just 16% of engineering roles and 26% of computing positions in the United States (NCWIT, 2024). At the leadership level, the gap narrows further: women account for only 4% of chief technology officers and have founded just 5.4% of tech startups (Silbert et al., 2025; World Economic Forum, 2024). The consequences are particularly acute in artificial intelligence where women author fewer than 14% of research papers and occupy only 18% of machine learning engineering roles (WomenHack, 2026). These disparities are often framed as pipeline issues tied to access and opportunity. However, the consequences extend far beyond classrooms and boardrooms, as women’s absence from technological design continues to shape and perpetuate technology-facilitated gender-based violence (TFGBV).
This is well documented by Caroline Criado Perez’s Invisible Women: Data Bias in a World Designed for Men (2019), which remains one of the most comprehensive accounts of what happens when women are systematically absent from the design process. Perez demonstrates that the designed world – from car seatbelts calibrated to male body proportions to voice recognition software trained on male voices – has been built around a male “default” so entrenched it has become effectively invisible. The design failures she documents were not necessarily malicious; they were structural. Because women were not in the room, they were not in the assumptions.
What Perez established for the physical world, the past decade of technology development has confirmed for the digital one – with considerably higher stakes. When a medical device is poorly calibrated for women’s bodies, the harm is individual and traceable. When an AI system is trained on skewed data, or consumer hardware is designed without women’s threat models in mind, the harm can escalate into Technology-Facilitated Gender-Based Violence (TFGBV), in ways that are far harder to attribute or correct. In parallel, a counter-movement has emerged: designers reclaiming the tools of engineering to build technology that centers the lived experiences of those most vulnerable to its failures. This shift proves that the “neutrality” of technology was always a myth; rather, technical design is a powerful tool of governance that, in excluding women, has produced the conditions for TFGBV – and in centering them, provides the infrastructure for liberation. The following sections examine both trajectories.
When Consumer Technology Becomes Infrastructure for Violence
When Apple launched AirTags in April 2021, the product was positioned as an elegant solution to a universal problem: losing inanimate objects. However, within months of launch, reports emerged of AirTags being utilized as high-precision tools for stalking. By 2024, documented cases had risen by 317% (Hughes, 2024), with law enforcement reporting a surge in complaints from women who discovered trackers hidden in bags, cars, and coat pockets. The data reveals a clear gendered pattern: 78% of stalking victims are female (MoralIt, 2025), and the majority of perpetrators in AirTag-related cases are people they already know. This distinction reveals a fundamental failure in Apple’s threat modeling. The company designed AirTags with the assumption that the primary risk was an unknown person planting a tracker. They did not adequately account for the far more statistically common scenario of domestic abuse, where the perpetrator already has access to the victim’s home or car (Wilson, 2021).
In September 2024, two Harvard students demonstrated that Meta’s Ray-Ban smart glasses could be paired with facial recognition software to identify strangers in real time. Their I-XRAY system could return a person’s home address and phone number within seconds of a simple glance (Harvard University, 2025). The demonstration highlighted that always-on cameras and large public databases create conditions for surveillance that are effectively invisible to their subjects. Meta’s glasses were, by design, the perfect vehicle for this invisibility. Indistinguishable from standard frames, their lack of a reliable recording signal was marketed as an aesthetic achievement. The consequences for women were immediate. By late 2025, reports of covert filming in public spaces surged, leading institutions to issue formal warnings about men using the glasses to record women on campus without consent (Schelij, 2025). While Meta included a small LED indicator as a nominal nod to transparency, the design was so easily bypassed with third-party modifications that it offered no true protection (Cox & Koebler, 2025).
The deepfake crisis is perhaps the most unambiguous in its gendered character. 98% of all deepfake videos online are non-consensual intimate content and 99% of the individuals targeted are women (UN Women, 2025). Between 2019 and 2023, the total volume of deepfake content saw a staggering 550% increase (UN Women, 2025). These are not marginal figures; they describe a technology whose primary real-world application, at scale, is the non-consensual sexual exploitation of women. The tools enabling this are widely available, largely free, and require minimal technical expertise (UN Women, 2026). Developed overwhelmingly by male engineers, the specificity of the harm these tools enable is itself a design artifact: many commercially circulating deepfake tools overwhelmingly target women and are optimized around female image datasets (UN Women, 2025).
The legal response has also struggled to keep pace with this architectural bias. The UK’s Online Safety Act, for instance, prohibits sharing manipulated explicit images but does not address their creation, nor the complexity of proving intent (UN Women, 2026). The burden of response falls almost entirely on the women targeted, who must navigate reporting systems and platform policies that – much like the technology itself – were not designed with their threat models in mind. The cases of AirTags, Meta Smart Glasses, and deepfakes are not isolated product failures. They are a structural pattern: the predictable outcome of designing technology without accounting for the people most vulnerable to its misuse. And while that pattern was taking shape, a quieter, more distributed counter-movement was building something different.
Reclaiming Technology Through Feminist Design
Moving away from venture-backed offices in San Francisco and corporate labs in Seattle, women have begun constructing their own computing devices from scratch at home. These builds are striking: functional computing hardware installed into pearlescent mini clutches, Hello Kitty makeup cases, and nostalgic 1990s toys (Steiber, 2026). The builds promote tech literacy, hands-on making, and individuality in an era where AI can homogenize creative work, attracting a predominantly female audience interested in non-traditional tech aesthetics (Rogers, 2026). Spearheaded by women technologists online, the cyberdeck movement is a creative call to action to take back control of how, when, and why we use technology (ARTECHOUSE, 2026). This movement finds its roots in 1980s cyberpunk fiction, where the cyberdeck was a portable, customized tool built for autonomy and resistance to corporate infrastructure (Steiber, 2026). Similarly, its contemporary version rejects the male “default” by reclaiming the tools of engineering – the hyper-feminine cyberdeck is a legible rejection of neutrality. It is a transition from being the subject of design to the author of it – transforming the computer from a black-box tool of surveillance into a bespoke instrument of autonomy, safety, and joy. It is from this cultural and political context that a new generation of women-led startups has emerged as its natural extension.
Currently in early development, Computer Angel prototypes a hair clip with a built-in camera. Unlike Meta’s smart glasses, which celebrated the technical achievement of invisibility, Computer Angel’s camera is deliberately, conspicuously large. This is an architectural argument: invisibility is a choice that prioritizes the wearer’s comfort over the subject’s awareness (Computer Angel, 2026). Computer Angel inverts this, using visibility as a deterrent. Research suggests that a visible camera reduces the likelihood of an attack by approximately 70% – a logic explicitly mirrored by the London-based startup eNOugh and its palm-sized “eNO badge” (Every, 2025). For these designers, transparency is the primary feature, not a design compromise. Understood within the framework of TFGBV, this design philosophy represents more than an aesthetic or commercial choice, it is a structural intervention. Where the tools created with the male “default” enabled violence by encoding invisibility, these products encode visibility, consent, and the threat models of their users into their architecture from the outset.
This ecosystem – including ROAR for Good’s Athena, Revolar, and Epowar – stands in direct contrast to the male “default” assumptions embedded in products like Apple AirTag. Designed by Yasmine Mustafa, Athena includes a silent-alert mode specifically intended for intimate partner violence situations, where an audible alarm could escalate danger rather than prevent it (ROAR, 2017). Revolar, developed by founders with personal experiences of family violence, sent more than 6,400 emergency alerts in its first year alone, demonstrating the significant demand for safety tools grounded in lived experience (Mannion, 2018). Similarly, Epowar incorporates biometric signals to trigger alerts automatically, recognizing a reality well documented in trauma research: during moments of acute threat, victims may be physically unable to press a button or activate a device manually (Words & Pixels, 2025). Each of these design decisions addresses a specific vector of TFGBV that existing consumer technology either ignored or actively enabled. Taken together, these products describe a coherent design philosophy – one rooted not in a shared institutional framework but in a shared experience of moving through a world whose technology was not built with you in mind, producing a consistent set of principles: the device cannot be turned against its user, its recording capacity is legible rather than covert, it accounts for the intimate partner threat model, and it is designed to function under conditions of stress and fear.
Designing Safety Into the Future of AI
The urgency of this shift cannot be overstated, as the industry moves from consumer hardware to the foundational infrastructure of AI. The consequences of low female-representation are already measurable: 44% of AI systems exhibit gender bias, and 25% exhibit simultaneous racial and gender bias (UN Women, 2024).
Because AI is trained on historical data, it risks becoming a “bias-automation machine,” encoding the exclusions of the past into the hiring, lending, and policing algorithms of the future (Goodman, 2018). The mechanism through which this bias reproduces itself is self-reinforcing. AI systems are trained on data, and that data reflects existing social inequalities. A hiring algorithm trained on past recruitment decisions will reproduce the gender biases embedded within them (Goodman, 2018). Facial recognition systems trained primarily on male faces perform less accurately on women, while content moderation systems often fail to recognize the forms of harassment women experience online (Buolamwini, 2018; Mauro & Schellmenn, 2023). In each case, the bias originates not at deployment, but during design – in the selection of training data, the definition of success metrics, and the assumptions shaping the system itself. Those decisions are ultimately made by the people building the technology.
Ultimately, design is never neutral. It is the product of specific people asking specific questions. As Caroline Criado Perez documented in the physical world, the digital world is currently being built on the foundations of the male “default.” However, the women building mermaid computers and angel cameras prove that a different trajectory is possible. As sociopolitical blogger Rachel Lowenstein said, “The same tools that build a prison can also build a cathedral.” Products like AirTags, Deepfakes and Meta’s smart glasses are not isolated design failures. They represent a pattern of TFGBV enablement, produced not by bad actors, but by the structural exclusion of women from the design process itself. Technology does not arrive with its uses predetermined; it is shaped by the hands that make it. Who builds determines who bears the risk. The window to change who builds is open. Whether the industry, its investors, and its regulators choose to treat that as a priority, before the assumptions of this generation’s AI become the unexamined foundations of the next, remains to be seen.
Interested in Working with Us?
If you’re interested in collaborating with Data-Pop Alliance on work addressing Technology-Facilitated Gender-Based Violence (TF-GBV), we welcome the opportunity to connect. Our projects combine ethical, participatory, and data-driven approaches to better understand how digital harms affect women, girls, and marginalized communities across different contexts. From online political violence and platform accountability to digital safety, care systems, and inclusive technology governance, DPA has experience designing research and policy-oriented initiatives that center lived realities and support evidence-based action.
Recent work includes projects such as Political Parties’ Responses to Technology-Facilitated Gender-Based Violence in Brazil, which examines how political actors and institutions respond to online violence targeting women in politics and public life. Across our portfolio, we aim to generate actionable insights that strengthen prevention, protection, and accountability efforts while amplifying underrepresented voices in digital spaces.
If your organization is exploring partnerships, research initiatives, or policy work related to TF-GBV, digital rights, gender equity, or inclusive technology governance, feel free to reach out to us at contact@datapopalliance.org with information about your goals, target communities, and areas of interest. Together, we can leverage data and collaborative research to support safer, more equitable digital environments.
About the Author: Zainab Ali is a Project and Research Officer living in Saudi Arabia.
![M002 - Feature Blog Post [WEB]](https://datapopalliance.org/wp-content/uploads/2025/10/M002-Feature-Blog-Post-WEB.png)
