International organizations funding Digital Public Infrastructure (DPI) programs in post-conflict societies risk creating surveillance systems while pursuing economic inclusion. Evidence from Sri Lanka and other comparative cases suggests donors must condition DPI support on verifiable rights protections, including multilingual communications, consultations, and legal safeguards. Such considerations should be pre-requisites to program funding, not merely governance afterthoughts.
Governments emerging from authoritarianism and protracted violent conflict face immense pressure to build administrative capacity quickly. Digital Public Infrastructure (DPI) promises efficient tax collection, rapid service delivery, and reduced corruption. International donors, viewing DPI as foundational to development, often commit substantial funding to such programs. Yet a context-blind approach risks a dangerous pattern: deploying powerful surveillance architectures before establishing the legal safeguards, transparency mechanisms, and public trust that make such systems governable.
This tension is clear across a range of fragile contexts. Colombia’s digital reparations registries struggled with duplicate claims and political contestation precisely because they lacked human rights-based design principles from the outset. Similarly, Uganda’s 2025 Express Penalty Scheme collapsed within months when citizens discovered vehicle fine systems operating without transparent data access policies.
These failures share a common structure: technically capable systems built atop foundations of institutional distrust. Brazil’s experience between 2020 and 2024 offers particularly instructive warnings. The federal government rolled out ‘Gov.br’, a unified platform providing access to over 4,000 public services, and ‘Pix’, an instant payment system now processing billions of transactions monthly. Both represented significant state capacity achievements. Yet when the Federal Revenue Office enacted a new norm in September 2024 to expand financial oversight, a communications vacuum allowed disinformation to metastasize.
In total, there were 1,770 fraudulent social media posts during January 2025 alone, with 949 specifically invoking the ‘Gov.br’ brand to perpetrate scams. A survey found 87% of Brazilians were exposed to the misinformation. Public authorities responded with rigid official communications and limited outreach, ultimately forcing the government to revoke the oversight measure on 15 January.
Technocrats designed sophisticated systems but invested little in the communications infrastructure needed to build understanding and consent. Operating at distance from end users, they created what researchers term an ‘orientation problem’: DPI was placed in social space without adequate consideration of how and why that space functions. The resulting vacuum allowed disinformation to flourish, eroding adoption and undermining promised efficiency gains.
Sri Lanka’s Digitalization Gamble
Sri Lanka appears poised to replicate this trajectory. For context, the 2026 budget allocates over 30 billion Sri Lankan rupees (about USD100 million) to digitalization initiatives centered on biometric identity systems, expanded government databases (linking revenue, immigration, police, and financial information), and a unified digital services platform. All sensible ideas at first glance.
Recent consultations with 60 officials, teachers, and activists revealed a striking gap in awareness. Despite being one of the state’s largest technology investments, the digitalization program was unfamiliar to all participants, including media-savvy professionals active across multiple news sources.
This communications vacuum becomes even more concerning when paired with structural and institutional racism. An analysis of Cyclone Ditwah’s official disaster communications documented systematic delays in Tamil-language updates from the Disaster Management Centre and Meteorology Department. Life-saving information in Tamil was never posted on social media. Evacuation orders reached Tamil fishing communities 12 hours behind Sinhala versions. Misinformation circulating in Tamil also went uncorrected. English-only place names rendered warnings unintelligible.
These are not isolated implementation glitches but evidence of how discrimination becomes encoded in infrastructure design when DPI systems are built without meaningful civil society participation. Indeed, not a single word in the DPI system design is made available in Sinhala or Tamil.
Dr Vagisha Gunasekara, a UNDP economist, warned that Sri Lanka’s approach risks “hard-wiring inequalities into the country’s economic backbone”. When a collective letter from over sixty civil society organizations went unanswered by the government, it implies a strategic choice to sideline expertise outside decision-making circles.
Prioritizing safeguarding: a better approach to DPI
The UN Universal DPI Safeguards Framework emerged precisely to address these recurring failures. Its 18 principles mandate clear legal bases for data collection, transparent multi-stakeholder governance to prevent ‘function creep’, independent oversight mechanisms, and purpose limitation rules constraining how information gathered for one service can be repurposed.
The framework recognizes that post-war contexts carry heightened surveillance risks. AccessNow’s human rights-centred approach similarly emphasizes that marginalized populations, ethnic minorities, political dissidents, and those previously targeted by state violence experience asymmetric distrust that shapes whether and how they engage with digital systems.
Yet economic sequencing arguments often sideline these safeguards as obstacles to rapid deployment. Donor organizations under pressure to demonstrate development impact accept promises that rights protections will follow once systems are operational.
However, research on adoption patterns in surveillance-prone contexts reveals what scholars term the ‘opt-out paradox’: people use systems not because they trust them but because exclusion from essential services leaves no alternative. For example, Jordan’s refugee populations and India’s social protection beneficiaries adopt biometric systems despite documented surveillance abuses because non-participation means forfeiting food assistance or legal recognition. This coerced adoption does not equal legitimate state capacity. It creates brittle systems vulnerable to a collapse in trust when abuses surface or when disinformation exploits the legitimacy deficit.
What next for policymakers and other stakeholders?
The policy intervention requires inverting current practice. International development institutions should condition DPI disbursements on verifiable benchmarks achieved before deployment. This could include publication of complete technical specifications in all official languages, mandatory human rights impact assessments with public findings, establishment of independent oversight bodies with enforcement powers, and documented civil society consultation processes with evidence of how feedback shaped design choices.
Brazil’s trajectory demonstrates that retrofitting trust mechanisms onto operational systems fails. Sri Lanka’s massive investment will only succeed if and when safeguards are treated as prerequisites rather than afterthoughts. Even so, it is unclear and uncertain if the government will adopt this approach.
The Sri Lankan government should recognize that its new digital systems require strong human rights guardrails from the outset. Evidence from fragile contexts worldwide suggests that without this reorientation, international donors risk their financing and the state risks creating an ‘efficient authoritarianism’ around pervasive DPI mechanisms that serve technocratic priorities, while systematically discriminating against the populations it claims to serve.
This article is part of a series produced in collaboration with the International Centre for Tax and Development at the Institute of Development Studies (IDS), UK, exploring the role of Digital Public Infrastructure (DPI) in strengthening state capacity and fostering development. We welcome contributions to this series.