Tech Workers' Resistance Guide
- Tech Workers' Resistance is a sociotechnical framework combining strategies like unionization, anti-surveillance practices, and collective action to challenge corporate control.
- It demonstrates a shift from individualized occupational activism to sustained labor organizing, empowering a diverse range of tech professionals.
- The guide evaluates various tactics, balancing impact and risk, and highlights methods from public resignations to building autonomous infrastructures.
“Tech Workers’ Guide to Resistance” denotes a body of strategy, practice, and theory concerning how workers in the computing industry contest harmful products, managerial control, surveillance, precarity, and the political uses of digital infrastructure. In the recent literature, resistance is not limited to whistleblowing or public protest. It includes unionization, open letters, lawsuits, leaking, quitting, mutual aid, privacy practices, anti-surveillance design, worker-led training, refusal to build, and the construction of infrastructures oriented toward autonomy, sovereignty, and collective safety rather than market competition alone (Abdalla, 11 Aug 2025, Guirat et al., 7 Aug 2025, Luka et al., 2021). The topic sits at the intersection of CSCW, HCI, labor studies, privacy engineering, AI governance, STS, and social movement theory, and it is increasingly framed as a sociotechnical rather than merely ethical problem.
1. Conceptual foundations
A central conceptual shift in the literature is the move away from narrow “tech ethics” and from what “Resistance Technologies: Moving Beyond Alternative Designs” calls “alternative designs.” Ben Green characterizes tech ethics as a “terrain of contestation” in which the issue is not whether ethics is desirable, but who defines it, how it is operationalized, and whether it has any force against corporate incentives. In that account, ethics initiatives are often vague and toothless, overly focused on individual engineers, and vulnerable to “ethics-washing,” in which firms adopt ethical language without altering underlying practice (Green, 2021). “Resistance Technologies” advances a parallel critique of sustainable-tech discourse: green alternatives, dashboards, and AI optimization are described as insufficient when they remain bounded by market logic and capitalist imperatives. The proposed alternative is “Resistance Technologies,” defined as systems useful “in times of climate crisis, war, colonialism, growing inequalities and power imbalance, and resource scarcity,” with anti-surveillance treated as a foundational component of sovereignty rather than an optional feature (Guirat et al., 7 Aug 2025).
This literature therefore recasts resistance as a question of political economy, infrastructure, and power. “Resistance Technologies” emphasizes autonomy and sovereignty, anti-surveillance and privacy, decentralization with no single points of failure, critical refusal, and solidarity and community focus. The paper explicitly situates these design values against dependency on infrastructures such as AWS or Starlink that can be withdrawn or weaponized during crisis, and against digital colonialism rooted in infrastructures, platforms, and protocols controlled by dominant corporate or state actors (Guirat et al., 7 Aug 2025). The same orientation appears in sociotechnical critiques of AI governance, where workers are said to possess standing to intervene through three claims to jurisdiction: subjection, control over the product of labor, and proximate knowledge of systems (Luka et al., 2021).
A second foundational move is from individualized morality to collective power. The AI governance literature models harm reporting in three stages—identification, governance decision, and response—and shows that worker action becomes decisive when management refuses consensus or suppresses internal critique. In that framing, harms are not reducible to technical negligence; they arise in complex social systems despite technical reliability, and workers occupy a unique role in identifying and mitigating them because of their direct experience, technical knowledge, and relationship to the products they build (Luka et al., 2021). This suggests that a guide to resistance is best understood not as a code of personal virtue but as a repertoire for contesting governance failures.
2. From occupational activism to labor activism
The recent wave of tech worker resistance is historically specific but not historically unprecedented. Research on Google, Microsoft, and other firms argues that contemporary tech activism is “comparatively moderate” relative to earlier labor and social movements, even though it is often portrayed as radical. Historical precedents identified in the literature include Polaroid workers’ anti-apartheid actions and Microsoft class-action lawsuits in the 1990s, while contemporary action clusters around Big Tech’s military contracts, workplace discrimination, layoffs, and algorithmic labor control (Abdalla, 11 Aug 2025). The broader claim is that technical workers are not apolitical anomalies but participants in a longer labor history.
Empirically, one of the clearest findings is a transition from occupational activism to labor activism. “Unlikely Organizers” reports that the number of reported collective actions by U.S.-based knowledge tech workers grew nearly tenfold, from 4 in 2017 to 46 in 2019, and then remained at a consistently high level. The composition of that activity also changed: labor activism accounted for 10% of actions in 2018, 41% in 2020, and 75% in 2022; when blended actions are included, the proportion rises from 20% in 2018 to 85% in 2022 (Tan et al., 2023). The paper argues that periods of intense workplace social activism preceded later periods of heightened labor activism, extending Fantasia’s “cultures of solidarity” argument to professional workers.
The mechanism proposed is iterative. Workers first engage in occupational activism around socially beneficial work, company mission, or external harms. Conflict with management and retaliation then produces deeper worker identification and more durable organizing. Open letters, media campaigns, and public ethical appeals thus function as low-threshold entry points, but they often become on-ramps to NLRB action, union drives, and broader labor campaigns when workers encounter firings, demotions, or procedural obstruction (Tan et al., 2023). The Google trajectory from Project Maven and Project Dragonfly to later labor organizing is treated as paradigmatic rather than exceptional (Abdalla, 11 Aug 2025).
Recent studies of organizers broaden the social base of this movement. “The Future of Tech Labor” emphasizes that the relevant constituency includes engineers, product managers, customer support specialists, QA analysts, logistics workers, gig workers, and union staff organizers. High-profile cases include Kickstarter, HCL America contractors, the Alphabet Workers Union, Amazon JFK8, and United Videogame Workers. Against narratives of prestige and privilege, the paper stresses fragmented and unstable work environments, workplace stratification, platformization, and surveillance as conditions that both motivate and impede organizing (Sum et al., 18 Aug 2025). Organizers respond by building community across silos, conducting one-on-ones, and using spaces such as Tech Workers Coalition and Labor Notes to create durable movement infrastructure.
3. Repertoires of action and strategic calculus
The most systematic recent attempt to classify tech-worker tactics is “$100,000 or the Robot Gets it! Tech Workers’ Resistance Guide,” which catalogues actions and evaluates their perceived risks and impacts through Best-Worst Scaling. In that study, actions analyzed in newspapers included whistleblowing and leaking (22%), internal passive protests (22%), lawsuits (16%), public protests (8%), unionization, disruptive workplace protests, public resignations, and others. The most common motivations were worker environment (28%), diversity and identity issues (16%), anti-war and Palestine protests (14%), and sexual harassment (10%). Split-half reliability for the expert annotations was high, with a Spearman correlation of 0.92 (Abdalla, 11 Aug 2025).
The paper’s central finding is that perceived impact and perceived risk are strongly correlated, but not perfectly so. Striking, disruptive protests, sabotage, and public leaking are characterized as among the most impactful and risky actions. Violence toward infrastructure is assigned $Impact = 0.825Risk = 0.975Impact = 0.525Risk = -0.3$. HR engagement, internal passive protests, low-key boycotts, and minimal productivity are classified as low-impact and low-risk (Abdalla, 11 Aug 2025). The larger strategic point is that no tactic is universally optimal; individual circumstance, immigration status, caregiving obligations, and organizational context matter.
| Action family | Examples named in the literature | Reported characterization |
|---|---|---|
| Quitting | Quiet quitting, quitting in protest, public resignations | Public action can increase impact; private action can reduce exposure |
| Protesting | Internal active/passive protests, public protests, disruptive workplace protests, striking | Striking and disruptive protests are among the most impactful and risky |
| Knowledge transfer | Anonymous or public leaking, publicizing actions, teach-ins, campaigns | Public leaking is high-impact and high-risk |
| Lawfare | Lawsuits, reports to regulators, HR engagement, lobbying | HR engagement is low-impact and low-risk |
| Positive and constructive actions | Building competing products, reparative actions, organizing, community-building | Building competing products is low-risk and moderate-impact |
A controversial extension of this discussion is the “radical flank.” Drawing on social movement theory, the guide argues that the absence of a true radical flank suggests tech worker activism remains a movement “in its infancy,” and that a more radical faction can make moderates appear more reasonable and thereby increase concessions to moderate demands. At the same time, the paper explicitly warns of a “negative radical flank effect” if the radical edge becomes too extreme or violent (Abdalla, 11 Aug 2025). This debate reappears in work on protestware, where software itself becomes a vehicle of protest. “An Investigation into Protestware” analyzed 32 samples and found three recurring characteristics: diversity in the “nature of inducing protest,” discriminatory “nature of targeting users,” and a “nature of transparency” that is not always respected. It also reports that disruptive protestware can cause substantial adverse impact on downstream users and that usage of protestware from JavaScript libraries has generally increased over time (Finken et al., 2024). The protestware literature therefore treats digitally embedded resistance as real, but ethically and infrastructurally fraught.
4. Surveillance, algorithmic management, and privacy
A large portion of the resistance literature concerns surveillance and metricized labor control rather than product ethics narrowly construed. “It’s Always a Losing Game” describes workplace surveillance technologies as layered systems spanning cameras, audio, AI, keyloggers, emotional recognition, productivity tracking, location monitoring, biometrics, computer activity, and personal data collection. Workers report privacy concerns, lack of transparency, health and wellbeing risks, and a “culture of distrust,” especially in remote and hybrid settings (Sum et al., 2024). Resistance in this setting is often everyday rather than spectacular: commiseration in Reddit and WhatsApp groups, information seeking, technological obfuscation such as mouse jigglers and separate user profiles, ignoring digital cues, “soldiering,” and quitting.
Research on Amazon fulfillment centers shows a related but more formally theorized pattern. “Fulfillment of the Work Games” situates worker tactics under algorithmic management as “work games,” adapting Burawoy’s framework to quantified, surveilled warehouses. Reported tactics include cherry-picking tasks, “virtual picking,” reordering work to game metrics, using the “menu screen trick” to pause timers, exploiting scan-to-scan loopholes, collaborative coverage to avoid “time off task,” makeshift chairs, passive resistance, and non-consenting resistance such as fake GPS for clock-in (Cheon et al., 13 Aug 2025). The paper’s terms “quantified objectification,” “agentic configuration,” and the dialectic of resistance and consent are important here: resistance may restore a measure of autonomy while still operating within, and sometimes reinforcing, productivity logics.
For organizers, privacy becomes both an individual and collective problem. “Weaving Privacy and Power” reports that labor organizers use a blend of digital security practices and community-based mechanisms. Organizers avoid company-controlled infrastructure where possible, migrate to personal devices and third-party tools, use Signal, Keybase, and Cryptpad for sensitive communication, manage access controls, and reduce digital paper trails through ephemeral communication. At the same time, they rely on moderation, explicit social norms, trust-building, and “safety in numbers,” since collective efficacy often requires going public together rather than maintaining perfect anonymity (Kapoor et al., 2022). The paper’s core argument is that individual privacy and collective privacy are intertwined, sometimes in conflict and often mutually constitutive.
Employer counter-organizing intensifies these pressures. “Surveillance, Spacing, Screaming and Scabbing” identifies four recurring technological tactics in union busting: surveillance, spacing, screaming, and scabbing. Surveillance uses data asymmetries to anticipate organizing; spacing isolates workers physically and digitally; screaming saturates required channels with anti-union propaganda; scabbing uses digital hiring, workflow rerouting, and centrally controlled knowledge systems to replace labor during strikes (Reiber et al., 3 Mar 2026). Across these studies, the implication is direct: resistance to digital labor control requires not only tactics of evasion but worker-controlled communication, collective privacy, and technical knowledge of the workplace infrastructure itself.
5. Sovereignty, solidarity, and transnational resistance
The resistance literature increasingly moves beyond the firm and beyond the Global North. Brazilian tech workers are presented as operating at the intersection of labor and technology within a context marked by colonial histories, socio-economic inequality, technological dependence, and underrepresentation in shaping digital futures. Their resistant practices include challenging “data universalism,” critiquing technosolutionism and the “California ideology,” resisting harmful platform metrics, pushing back on content moderation bias, advocating for AI trained on Brazilian Portuguese and local geospatial and visual data, and building grassroots initiatives such as the MTST Technology Sector (Seto, 9 Jan 2026). Examples include platforms for delivery cooperatives such as Señoritas Courier and mobile apps for illiterate women in extractivist communities that use audio and visual interfaces and analogies to local knowledge.
This orientation is closely aligned with the sovereignty thesis of “Resistance Technologies.” There, sovereignty is not a militarized or nationalist term but one of technical self-determination and community empowerment. Anti-surveillance technologies are treated as foundational for resistance against digital colonialism, gendered abuse, and infrastructures controlled by powerful states or corporations (Guirat et al., 7 Aug 2025). A plausible implication is that for tech workers, resistance is not exhausted by opposing particular contracts; it also includes reducing infrastructural dependency and designing for conditions of polycrisis.
Community-centered safety appears in adjacent activist research. A survey of 50 Black Lives Matter activists found that 84% used social media to find information, 64% had concerns about data privacy or surveillance on social media, 40% had concerns about the credibility or reliability of information online, and 72% reported that personal networks helped them find information or feel safe attending protests (Rosenbloom, 2022). The requested technological properties—reduced corporate involvement, stronger privacy and security features, end-to-end encryption, anonymity, better authentication, and community-controlled platforms—map closely onto the anti-surveillance, decentralization, and solidarity principles articulated in “Resistance Technologies” (Guirat et al., 7 Aug 2025).
The solidarity horizon also extends to carceral systems and Indigenous data governance. “Examining Solidarity Against AI-Enabled Surveillance at the Intersection of Workplace and Carceral Realities” argues that workplace and carceral surveillance infrastructures overlap through tools such as facial recognition, CLEAR, and Veriato, and that resistance requires alliances between workers and people impacted by the carceral system (McErlean et al., 8 Oct 2025). “The Role of Social Movements, Coalitions, and Workers in Resisting Harmful Artificial Intelligence” adds Indigenous data sovereignty, the CARE principles, algorithmic impact assessments, and coalitions such as Athena and Mijente to this wider field of resistance, emphasizing that worker organizing is strongest when linked to racial justice, immigration justice, housing, climate, and anti-surveillance movements rather than confined to corporate policy debate alone (Struensee, 2021).
6. Organizational supports, counter-institutions, and skill formation
Resistance is not only oppositional. Several papers treat it as a matter of building durable capacities inside and around organizations. “Knowledge Workers’ Perspectives on AI Training for Responsible AI Use” reports that workers often receive inadequate, fragmented, and management-led training. In response, participants surfaced nine training topics: foundational understanding, exposure to tools and capabilities, literacy and critical thinking refreshers, critical examination and validation, DEI harms and benefits, dataset representation and biases, worker rights and AI, grievance mechanisms, and worker data privacy (Zhang et al., 8 Mar 2025). The paper’s emphasis on worker rights, contestation procedures, and participatory design shifts training from compliance to empowerment.
A parallel argument appears in work on human resilience. “Human Resilience in the AI Era” defines resilience across three layers: psychological, social, and organizational. Emotion regulation, meaning-making, and cognitive flexibility operate at the individual level; trust, social capital, and coordinated response operate at the group level; psychological safety, feedback mechanisms, and graceful degradation operate at the organizational level (Liu et al., 29 Oct 2025). The paper explicitly states that resilience can be cultivated through training that complements rather than substitutes for structural safeguards. In this context, resistance is not simply refusal; it is also the preservation of agency under AI-mediated strain, silent failure, and digital overload.
Institutional resistance also includes the maintenance of inclusive practices under backlash. “From Diverse Origins to a DEI Crisis” documents several adaptive strategies in software engineering: maintaining ERGs, preserving visible symbols and messaging, leadership statements affirming diversity and inclusion, rebranding DEI under other labels, strategic framing through performance or compliance language, and “quiet continuity” through less public but still active mentoring and support structures (Santos et al., 23 Apr 2025). The study treats rebranding, strategic framing, and quiet continuity not as abandonment but as calculated responses to a contested environment.
Finally, new AI modalities generate new sites of resistance. “How Tech Workers Contend with Hazards of Humanlikeness in Generative AI” describes an “unsettled knowledge environment” in which workers across ML engineering, policy, UX, product management, technical writing, and communications confront hazards such as additional labor, miscalibrated trust, believability and fraud, worsened UX and social interaction, disparate AI literacy, and worker replacement or skill erosion (Díaz et al., 22 Dec 2025). The paper recommends clearer conceptions of humanlikeness, explicit hazard mapping, role-specific guidance, and cross-functional coordination. Here again, the theme is infrastructural: resistance requires institutions that can transform situated worker knowledge into durable governance.
7. Persistent tensions and future directions
The literature presents resistance as necessary but internally contested. One recurring tension concerns the difference between individual workaround and collective transformation. Everyday tactics—virtual picking, mouse jigglers, separate user accounts, pseudonyms, quitting—demonstrate ingenuity, but several studies warn that they can remain isolating, risky, or partially complicit with managerial logics (Cheon et al., 13 Aug 2025, Sum et al., 2024). A second tension concerns internal versus external action. Open letters, ethics boards, and HR processes may create low-threshold entry points, yet research on tech ethics and AI governance repeatedly shows that internal channels are often subordinated to corporate interests and can become vehicles for ethics-washing or reputational management (Green, 2021, Luka et al., 2021).
Another major controversy concerns militancy. The radical-flank argument holds that a movement composed entirely of moderates may lack leverage, while the protestware literature demonstrates that technically disruptive tactics can impose real downstream harms, especially when targeting is discriminatory or transparency is absent (Abdalla, 11 Aug 2025, Finken et al., 2024). The empirical record does not resolve this dispute. It does, however, show that corporations adapt: they increase retaliation, alter contracts to prevent project cancellation under worker pressure, restructure communication channels, and build digitally mediated anti-union capacities (Abdalla, 11 Aug 2025, Reiber et al., 3 Mar 2026). Resistance therefore faces an adversary that learns.
The future directions proposed across these papers are comparatively consistent. They include worker-led design and policy, stronger transparency and grievance mechanisms, collective rather than purely individual privacy protections, sectoral and cross-company organizing, transnational solidarity, infrastructure for movement learning, and anti-surveillance systems designed for autonomy and crisis resilience (Kapoor et al., 2022, Sum et al., 18 Aug 2025, Reiber et al., 3 Mar 2026, Guirat et al., 7 Aug 2025). They also include alliances with communities most affected by digital harm, especially marginalized groups, workers under algorithmic management, and people subject to carceral surveillance (McErlean et al., 8 Oct 2025, Struensee, 2021). This suggests that a mature guide to resistance is neither a checklist nor a doctrine. It is a sociotechnical repertoire through which workers build power, contest harmful systems, and create institutions capable of sustaining autonomy, solidarity, and democratic control over computation.