• UN Study Highlights High Carbon Footprint of Hong Kong Data Centres

  Hong Kong's data centres are among the world's most carbon-intensive due to the city's heavy reliance on fossil fuels for energy. By 2030, global data centres could consume 945 terawatt-hours of energy, nearly triple the electricity usage of Pakistan, Bangladesh, and Nigeria combined. Environmental Impact Hong Kong's grid has a carbon footprint 43% higher than the global average, trailing only Indonesia and India. Energy sources in Hong Kong consist of 67% fossil fuels, 32% nuclear, and only 1% renewables. While the global AI infrastructure footprint includes massive water and land usage—potentially consuming water equivalent to the needs of 1.3 billion people—Hong Kong ranks lower in these specific metrics due to its lack of renewable energy reliance, which typically requires more land. AI Energy Consumption Approximately 80% to 90% of energy demand for AI is driven by day-to-day model usage rather than initial training. Image generation requires significantly more energy than text queries, with ChatGPT processing 2.5 billion prompts daily. Rapid growth in AI platforms, such as DeepSeek, has contributed to escalating demand for data processing power. Local Policy and Infrastructure Hong Kong remains a key hub for data centres with 300 internet service providers and extensive submarine cable connectivity. The government is expanding infrastructure, including a new 110,000-square-metre facility in Sandy Ridge. The Electrical and Mechanical Services Department (EMSD) currently manages energy efficiency via its Green Data Centres Practice Guide, which regulates cooling systems and design standards.

  Global Voices · Jun 9, 2026

  

• AI cameras begin surveillance of roads in Dhaka, Bangladesh

  Since May 7, 2026, the Dhaka Metropolitan Police (DMP) has deployed AI-powered traffic cameras at key intersections. The system automates traffic enforcement by filing cases without human intervention, sending notices to vehicle owners within hours of a violation. Over 1,000 traffic cases were filed in the first 10 days, with plans to expand to at least 500 locations across the capital within six months. How the System Operates AI cameras detect red-light running, stop-line crossing, and lane violations, automatically scanning license plates. The system cross-references data with the Bangladesh Road Transport Authority (BRTA) to issue penalties directly to the vehicle owner. Violations result in fines and the deduction of points from the owner's driving license, regardless of who was driving at the time. Criticisms and Challenges Selective Enforcement: Citizens have expressed frustration that while private cars are targeted, buses, unregistered vehicles, and rickshaws frequently commit violations without being held accountable. Invisible Vehicles: The system relies on license plates, meaning unregistered battery-powered rickshaws—a major source of congestion—remain largely invisible to the technology. Security and Fraud: Reports of scams have emerged where fraudsters call owners demanding money for fake fines; additionally, public confusion regarding camera maintenance (e.g., equipment fallen due to wind) has fueled misinformation. Expert Perspective Professor B M Mainul Hossain of the University of Dhaka emphasized that while technology is a step forward, long-term success requires system transparency, an appeals process, and a comprehensive policy for handling unregistered transport.

  Global Voices · Jun 3, 2026

  

• Artificial Intelligence Governance and Digital Rights in Ecuador

  Ecuador faces an uncoordinated landscape for AI policy, with various government and academic entities operating without a unified framework. A late 2025 gathering at FLACSO Ecuador brought together civil society, academia, and public institutions to address the intersection of AI and digital rights. Recent developments include the government's 2026 launch of the EFIA-EC strategy, alongside increased private sector engagement such as the arrival of Palantir. Key Challenges Privacy and Data Security: Public workers lack institutional tools, leading to sensitive data, such as medical information, being processed on personal AI accounts. Education: Teaching institutions struggle to balance the use of AI tools with concerns over plagiarism and the erosion of critical thinking skills. Inclusion and Bias: AI systems frequently rely on Western datasets, marginalizing Indigenous and local knowledge systems. Legislative and Strategic Landscape National Assembly: Three legislative bills from 2024 were unified, but progress has stalled with suggestions that the proposals may be shelved. Governance Vacuum: There is no consensus on whether to prioritize regulating AI use, development, investment, or automated decision-making. EFIA-EC Strategy: Released in March 2026, this four-year roadmap focuses on governance, capacity building, and responsible development with support from the IDB and UNESCO. External Influences Palantir Entry: President Daniel Noboa's engagement with Palantir CEO Alex Karp has led to the firm opening an office in Ecuador, raising concerns about surveillance and intelligence data management in the absence of robust regulation. UNESCO RAM Methodology: Used to assess institutional readiness rather than just technical capabilities, ensuring countries can sustain AI systems without compromising human rights.

  Global Voices · May 30, 2026

  

• How AI is Upgrading African Dictatorship

  Artificial intelligence (AI) is being leveraged by authoritarian regimes across Africa to enhance mass surveillance and suppress political dissent before it manifests. A March 2026 report indicates that 11 African governments have spent over USD 2 billion on AI-powered surveillance, ostensibly for public safety but primarily targeting opposition hubs, protests, and independent media. The Surveillance Upgrade Regimes are deploying Chinese- and Israeli-supplied technologies, including high-definition CCTV, plate readers, and facial recognition, to transform dormant administrative data into queryable surveillance archives. AI allows regimes to synthesize travel patterns, transaction histories, and digital footprints into a 'loyalty index' that can monitor citizens' perceived intentions rather than just their actions. Pre-emptive Suppression Modern digital surveillance operates as a panopticon: the goal is to make citizens aware they are being watched, thereby deterring dissent before it begins. Algorithms enable pre-emptive targeting; identifying an organizer through metadata often leads to their marginalization or intimidation without the need for physical police intervention or formal charges. Limitations and Implications While civic-tech groups are working to document abuses and push for rights-based governance, they face an insurmountable asymmetry in resources and infrastructure compared to state actors. Experts warn that AI is significantly lowering the cost and increasing the ceiling for authoritarian control, making it possible for regimes to monitor entire populations without the need for the massive informant networks required by 20th-century dictatorships.

  Global Voices · May 22, 2026

  

• From Blooms to Bytes: Kenya’s Data Center Boom and the Risk of Digital Colonialism

  Kenya faces a new industrial "scramble" as Microsoft and G42 construct a USD 1 billion data center in the Olkaria geothermal fields, mirroring the 1980s floral industry boom in Naivasha. Critics fear the project may repeat past patterns of exploitation, specifically regarding land rights, water consumption, labor abuses, and tax avoidance. The facility's "green" credentials are being questioned due to potential aquifer depletion and the intense water requirements of evaporative cooling systems. Economic Risks and Revenue Bypass The project risks creating an "economic enclave" that bypasses national revenue-generating systems: Power: By sourcing electricity directly from KenGen, the facility bypasses Kenya Power (KPLC), depriving the national distributor of industrial revenue and potentially forcing tariff hikes for ordinary citizens. Tax: The Kenya-UAE Double Taxation Agreement (DTA) may allow companies like G42 to shift profits to the UAE’s lower-tax jurisdiction, reducing the tax base for the Kenya Revenue Authority. Labor: Potential use of independent contractors and third-party outsourcing may allow the companies to avoid mandatory contributions to social health and housing funds. The "Green" Energy Paradox While the project uses renewable geothermal energy, its water footprint poses severe risks to the semi-arid Olkaria and Lake Naivasha basin. Data centers rely on evaporative cooling, which "consumes" water by converting it into steam. Excessive groundwater extraction threatens local wells, livestock, and small-scale farmers, leading to potential resource conflicts. Concerns persist regarding thermal pollution and chemical runoff from "blowdown" water returning to the environment.

  Global Voices · May 8, 2026

  

• On the need to generate more questions than pleasant answers: 'Don't ask AI, ask a peer'

  The importance of human knowledge sharing is being reclaimed over reliance on AI-generated, homogenized outputs. Global Voices, APC, and GenderIT collaborated on a series titled "Human perspectives on AI" to re-center human diversity, creativity, and lived reality in the face of algorithmic control. Challenging the Human-AI Dichotomy The concept of "human" has been weaponized by Western, patriarchal, and cisheteronormative power structures to define marginalized populations as "less human." Massive corporations like OpenAI collaborate with surveillance and military sectors, while decentralized civil society networks, such as Brazil's "Código Não Binário," build open-source tools like TybyrIA to combat hate speech. Expert Perspectives on AI Daria Dergacheva notes that generative AI is merely probabilistic automation, fundamentally distinct from human cognition and creativity. Kira Xonorika highlights that traditional, non-human, and ecological intelligences have been subordinated by a narrow, tech-centric view of "humanity." Hija Kamran emphasizes that essential human functions—care work, community building, and empathy—cannot be automated. Real-World Impacts and Violations Rebecca Ryakitimbo argues that human rights advocacy must be central to AI regulation to shift power back to the public. Rezwan details how facial recognition and surveillance tools restrict civic space for marginalized communities in India. Mariana Tamari reports on traditional communities in Brazil facing eviction through land-grabbing tactics enabled by unrestricted digitalization, noting that corporate tech narratives often ignore real-world climate and territorial collapse. Generating Questions Over Answers The series demonstrates that discomfort and debate are more valuable than the "palatable" but biased answers provided by AI. Challenging the notion that AI is "inevitable" requires identifying who creates these tools, why they are created, and which lived realities are ignored in the process. The call to action is to reject pre-programmed responses and embrace uncertainty by engaging in peer-to-peer discourse to reclaim agency and transparency.

  Global Voices · May 5, 2026

  

• Ethiopian creator Boni’s death sparks debate on platform accountability

  Boni, a popular Ethiopian content creator who produced videos in Afan Oromo, died by suicide on April 29, 2026. Before her death, she experienced recurring online harassment on TikTok, including content that violated the platform's hate speech policies. Despite posting warning signals on Facebook and TikTok, both platforms' automated moderation systems failed to detect or intervene in time. While TikTok removed her final videos post-mortem, they remain accessible through reposts and stitches, and Meta has kept her content live on Facebook. Platform Failures and Policy Gaps Both TikTok and Meta market robust, multilingual AI moderation as a core capability, yet these systems failed to flag clear indicators of crisis. TikTok’s failure to act on documented harassment—where other creators explicitly wished for her death and miscarriage—left a vulnerable user exposed. The "industrial hash-matching" technology, typically used to block copyrighted music or terrorism content, was not effectively applied to prevent the viral spread of Boni’s final moments. Meta’s refusal to remove the content from Facebook contradicts its own Community Standards regarding self-injury and suicide. The Misplaced Debate in Ethiopia Public discourse among Ethiopian creators has largely focused on "toxic" audience behavior, bullying, and jealousy, while ignoring the role of the platforms themselves. Platforms are not neutral conduits; they are active agents that choose which behaviors to monetize and amplify via their recommendation algorithms. The current debate overlooks the fact that platforms are actively harvesting attention in Ethiopia without investing in adequate, localized moderation pipelines. Resource for Support If you or someone you know is in crisis, help is available. You can find international support resources at Befrienders.org.

  Global Voices · May 5, 2026

  

• Global Perspectives on the Uneven Impact of AI

  The "Human perspectives on AI" spotlight highlights how artificial intelligence is distributed and experienced unevenly across the globe, impacting communities differently based on geography and socioeconomic status. Technologies and their by-products follow a pattern where high-profit concentration occurs in wealthy nations while labor-intensive and environmentally damaging infrastructure is often exported to lower-income regions. Information access remains skewed, with mainstream media focusing on limited regions, obscuring localized AI impacts in areas like Latin America, Asia, and Eastern Europe. Global AI Implementation and Risks Surveillance and Border Control: AI is being used for invasive surveillance in India and for both exclusionary and facilitative roles at different international borders. Economic Disparities: While wealthy nations see hype around agentic AI, other regions face cost-related disappointments or labor exploitation through low-paid moderation tasks. Language Preservation: There is ongoing debate regarding the potential and risks of AI in supporting or harming minority and Indigenous languages. Environmental and Resource Impacts Data Centers: The environmental burden of AI, including water consumption and energy waste, is being exported to lower-income countries. Regional Resistance: Local communities are increasingly pushing back against data center construction in Latin America and Asia, while regions like Singapore have implemented moratoriums due to water scarcity concerns. Socio-Political Implications Narratives and Hype: AI integration into systems like healthcare (e.g., in El Salvador) is often driven by specific promotional narratives rather than objective necessity. Human-Centric Alternatives: Collaborations like "Don’t ask AI, ask a peer" examine AI through human rights, feminist, and ethical lenses, questioning what we should actually mechanize. Cultural Expression: Examples like a Bosnian rock group's satirical song illustrate how AI affects culture and creative workers globally, providing a counter-narrative to corporate-led AI optimism.

  Global Voices · May 4, 2026

  

• The AI Access Divide: Infrastructure Realities in Sub-Saharan Africa

  Less than 12 percent of the population in Sub-Saharan Africa currently has the technical, economic, and linguistic requirements to access artificial intelligence. Persistent power outages and inadequate digital infrastructure serve as fundamental barriers to meaningful AI adoption. The gap between political rhetoric regarding national AI strategies and the on-the-ground reality of basic utility services remains profound. Key Barriers to Adoption The Energy Divide: AI infrastructure and end-user devices (smartphones, routers) are entirely dependent on stable electricity, which remains unavailable to millions. Connectivity Thresholds: While many users have 2G or 3G access, generative AI requires stable 4G/6G or fiber-optic connections, low latency, and high bandwidth. Linguistic Exclusion: Large Language Models (LLMs) often fail to recognize or support local African languages, such as Guin in Togo, making dialogue and communication impossible for many. The Problem with Stopgap Solutions Adaptations like offline models or lightweight AI are often funded for limited purposes; while they provide value in sectors like agriculture or education, they may inadvertently create a second-tier of access. Relying on these interim solutions risks shielding governments and corporations from the necessity of addressing structural issues like basic electrical grid stabilization. Stakeholder Responsibilities Governments: Must prioritize electrification, regulatory reform, and institutional stability before pursuing advanced technological feats. Telecom Operators: Need to reduce data tariffs, increase market competition, and invest in rural infrastructure to ensure network availability. AI Companies: Should move beyond superficial hub agreements with elites and instead develop multilingual models in genuine partnership with local stakeholders who understand practical constraints. Local Civil Society: Must continue developing SMART solutions tailored specifically to real-world local contexts and needs.

  Global Voices · May 4, 2026

  

• The Human Cost of the Data Center Boom in Latin America

  Data centers are expanding rapidly in the Global South, driven by AI demand, with projections of 20 percent annual growth until 2030. These facilities often operate with minimal regulatory oversight, consuming massive amounts of water, electricity, and land while straining local infrastructure. Researchers like Steven Gonzalez Monserrate describe this as "Terraforming," noting how tech giants act like states to secure land and resources, often entrenching existing inequalities. Local communities in Brazil, Chile, Uruguay, and Argentina are increasingly resisting these developments due to environmental concerns and lack of transparency. Environmental and Social Impacts Resource Scarcity: Data centers require consistent, massive supplies of water for cooling and power for operations, often diverting resources from local residents in areas already struggling with drought. Pollution: Facilities generate noise pollution, which is linked to stress and high blood pressure, and rely on diesel or gas backup generators that cause air pollution. Heat Islands: NASA data shows that large data centers can create "heat islands," measurably raising local temperatures above surrounding areas. Limited Job Creation: Contrary to political promises, these centers require few personnel once operational, providing minimal long-term economic benefit to local populations. Case Studies of Resistance Brazil: The "AI City" project in Eldorado do Sul—a city recently devastated by historic floods—was fast-tracked, exempting the company from Environmental Impact Studies. Chile: Residents in Cerrillos successfully petitioned an environmental court to halt a Google data center, forcing the company to reconsider water usage, though the project remains in limbo. Uruguay: Following massive public outcry, Google was forced to modify its plans to switch from water-based cooling to an air-conditioning-based system. Argentina: The government is incentivizing large-scale data centers like the "Stargate" project via tax breaks, despite lacking specific regulatory frameworks for their environmental or social impacts.

  Global Voices · Apr 29, 2026

  

• Lack of Electricity and Internet Access Hinders AI Adoption in Sub-Saharan Africa

  Frequent power outages and high internet costs pose fundamental barriers to AI adoption in Sub-Saharan Africa. Experts argue that national AI strategies are premature without first establishing stable energy and connectivity infrastructure. Electricity as a Critical Bottleneck As of 2025, approximately 600 million Africans lack electricity, with 85% of these individuals residing in Sub-Saharan Africa. Even in nations with better infrastructure like South Africa and Côte d'Ivoire, daily power cuts disrupt hospitals, businesses, and households. IT expert Folli Herbert Amouzougan emphasizes that prioritizing a stable electrical grid reaching at least 95% of the territory is a necessary foundation for any technological advancement, describing current AI political rhetoric as "getting ahead of ourselves." Internet Quality and Affordability While mobile coverage is expanding, low-quality bandwidth and high latency make real-time AI tools unusable for many. The cost of data is prohibitively high; in many countries, 1 GB of mobile data costs between 2% and 10% of the average monthly income, rendering cloud-based AI tools inaccessible for most. Practical Implications and Inequality Local startups and SMEs face significant competitive disadvantages, including high operational costs, the need for VPNs, and slow connection speeds when accessing global AI models. UNESCO data shows most primary and secondary schools in the region are offline, raising concerns that AI-powered education will only widen existing learning gaps between urban and rural students. Limitations of Current Solutions Off-grid solar energy solutions, such as those provided by M-Kopa and Bboxx, are insufficient for AI as they typically only power small appliances, lacking the capacity to run high-performance computers, Wi-Fi routers, or local servers required for AI. Meaningful AI progress requires prioritizing political and economic commitments to infrastructure, rather than relying solely on the announcements of global tech giants.

  Global Voices · Apr 28, 2026

  

• The Threat of AI Deepfakes and Digital Violence Against Women in Brazil

  Teenagers across Brazil—including in Rio de Janeiro, Bahia, and Mato Grosso—have been reported for creating and sharing AI-generated non-consensual sexual imagery of classmates. Independent research center Internetlab has released a technical note categorizing such deepfakes as an "excessive risk" and calling for urgent regulatory intervention. Data from Security Hero indicates that 98% of deepfake videos online are sexually explicit, with women comprising 99% of the targets. The Intersection of Gender Violence and AI Internetlab research director Clarice Tavares argues that online and offline violence are manifestations of the same systemic misogyny. AI models often rely on biased training data, which inherently mirrors and amplifies gender-based prejudices and violence. Embedded AI tools on social media platforms, such as X's Grok, are criticized for lowering the barrier to entry for the creation and widespread dissemination of harmful content. Brazil's Legal Landscape The Brazilian Supreme Federal Court (STF) recently ruled Article 19 of the 'Marco Civil da Internet'—the country's internet civil rights framework—partially unconstitutional, citing the need for greater platform accountability. Despite ongoing debates over legislative criminalization of misogynistic behavior, experts warn that relying solely on criminal law limits the efficacy of victim reparation and prevention strategies. Policy Recommendations Implementation of "safety by design" principles is recommended to prevent the generation of harmful content at the software conception phase. Promotion of digital literacy programs to foster critical analysis of algorithms and AI ecosystems. As Brazil approaches election cycles, concerns are growing regarding the use of AI tools like Gemini, ChatGPT, or Claude to propagate gender-based political violence.

  Global Voices · Apr 28, 2026

  

• Bosnian band Dubioza Kolektiv satirizes blind trust in artificial intelligence with new song "Yebiga"

  The Bosnian band Dubioza Kolektiv has released a new song, "Yebiga," which offers a sharp critique of society’s growing dependence on artificial intelligence (AI) and the rise of a technocracy. The music video employs characters and elements from the legendary Yugoslav satirical show "Top Lista Nadrealista" (TLN) to highlight the absurdity of surrendering human agency to automated systems. Through humor and accessible lyrics, the band encourages audiences to reflect on the dangers of outsourcing critical thinking to opaque algorithmic platforms. Technocracy and the Erosion of Thought The lyrics identify figures like Elon Musk, Peter Thiel, and Jeff Bezos as symbols of an emerging technocracy, replacing democratic processes with algorithmic control. The song ironically mocks the public's preference for convenience, with lyrics suggesting that because AI can think and make decisions, humans no longer need to use their brains or read books. The central refrain, "AI is the boss now; there’s no going back," emphasizes a sense of resignation toward technological dominance. Utilizing Cult Comedy as Satire Directed by Vedran Mujagić, the video features actor Zenit Đozić reprising his role as the character Đipalo Junuz from the influential 1980s show TLN. The video revives the "HEPEK," a fictional device from the original show that was used to resolve societal conflicts; in the music video, an updated version is used as an absurd solution to dilemmas created by AI. This reference connects the contemporary critique of AI to a rich history of Balkan political satire, effectively using cultural memes to expose flawed logic in modern technological adoption. Societal Impact Known for addressing issues ranging from corruption to conspiracy theories, the band uses its platform to challenge the regional challenges of media literacy and information integrity. By framing the complex implications of automation and AI in an accessible, humorous format, the band asks an urgent question regarding the consequences of losing individual critical thinking in an age of machine-driven decision-making.

  Global Voices · Apr 26, 2026

  

• Decolonizing AI at the U.S. Border

  The Black Alliance for Just Immigration (BAJI) and UC Irvine Law’s clinics argue that AI-driven U.S. border enforcement violates international human rights laws, specifically the International Convention on the Elimination of All Forms of Racial Discrimination (ICERD). Surveillance technologies like autonomous towers and drones mark migrants of color as security threats, leading to increased dangers rather than safety. Algorithmic tools like CBP One, the Automated Targeting System (ATS), and ICE predictive models suffer from severe racial bias, lack of transparency, and inadequate oversight. Activists propose a "decolonial" approach to AI based on the African framework of Cosmo uBuntu to ensure diverse human representation and accountability in development. Surveillance and Algorithmic Bias Border Surveillance: Systems like Anduril Towers and Small Unmanned Aircraft Systems (sUAS) replace human patrols, forcing migrants into more perilous terrain and resulting in higher death rates. Entry Processing: The CBP One app struggled to identify darker skin tones and lacked language support; the Automated Targeting System (ATS) disproportionately flags travelers from certain nations, such as Nigeria, as high-risk. Internal Enforcement: ICE utilizes private algorithms like the "Hurricane Score" from B.I. Incorporated and the RAVEn platform to track individuals, often relying on biased local and international law enforcement data. Application and Adjudication Risks Asylum Review: The USCIS Asylum Text Analytics (ATA) tool may unfairly penalize non-English speaking applicants whose narratives resemble others, potentially rejecting legitimate claims. Evidence Classification: The AI-powered Evidence Classifier reviews massive amounts of documentation; atypical records can trigger negative outcomes for vulnerable applicants. Policy Recommendations Transparency and Redress: Agencies must notify individuals when AI decisions are made and provide opt-out options. Legal Safeguards: New federal laws should prohibit AI use cases that exacerbate structural racism and mandate independent oversight. Civic Action: Municipalities are encouraged to refuse information sharing with DHS for AI development purposes until systems are proven non-discriminatory.

  Global Voices · Apr 25, 2026

  

• The Technological Transformation of Colombia's Armed Conflict

  Colombia's armed conflict is being reshaped by the integration of AI and digital technologies. Evolution ranges from the tactical use of low-cost drones by non-state actors to the state's deployment of complex anti-drone defense systems. AI-driven disinformation and digital community control methods are amplifying existing regional inequalities and patterns of violence. Battlefield Transformations and State Response Low-cost drone use: Since 2024, non-state armed groups have used modified commercial drones for explosive attacks on police and military targets, demonstrating how accessible technology can alter tactical power balances. National Anti-Drone Shield: Since 2025, the government has been building a comprehensive architecture utilizing specialized sensors, micro-Doppler identification, and radio frequency blocking to neutralize incoming threats. Algorithmic Risk Management Operational tools: Both police and military forces have adopted AI-driven systems to analyze data for risk modeling and crime hotspot identification. Risks of bias: Due to regional inequalities and poor data quality, there is a risk that these systems will reinforce stigmatization and intensify surveillance in specific areas while making others invisible. Manipulation and Social Control Disinformation: AI-generated audio of political candidates and deepfake avatars mimicking doctors or military personnel have been used to influence elections and incite protests since 2023. Digital blacklists: Non-state actors exploit closed WhatsApp groups and anonymous Facebook profiles to share profiles of community leaders, facilitating threats and targeted violence. AI threatens to automate and accelerate these control systems. Implications The immediate danger is not necessarily the rise of fully autonomous weapons, but the risk that these technologies reinforce and amplify the deep-seated social disparities that have fueled decades of conflict in Colombia.

  Global Voices · Apr 25, 2026

  

• Between the Algorithmisation of Territories and the Monoculture of Data: Paths Toward Rights-Based AI

  Digitalized agribusiness, driven by an alliance of Big Tech and Big Agro, threatens traditional land rights and erodes local knowledge. Algorithmic systems are institutionalizing "land grabbing" and displacing communities, while technologies like drones are used to intimidate smallholders. A human-rights-based approach to AI requires dismantling techno-solutionism and prioritizing decentralized, community-driven governance. The Myth of Precision Agriculture Industry narratives market "precision agriculture" and AI-driven automation as magical cures for environmental degradation and labor shortages caused by monoculture. This corporate-led digitalization reduces complex biomes and human-nature relationships into massive data sets, rendering traditional communities invisible to remote operators. Impacts in the Matopiba Region In Brazil’s Matopiba region, the Rural Environmental Registry (CAR) system is used to facilitate illegal land appropriation. Automated systems fail to distinguish between ancestral land occupation and recent invasions, allowing large landowners to register public or collective lands as private assets. Traditional communities, such as those in Gleba Tauá, face violent displacement as digital documentation legitimizes land grabbing and drones are deployed to spray agrochemicals near homes and water sources. Paths Toward Rights-Respecting AI The design and architecture of AI must be transparent and held under strict public and civil society scrutiny to move beyond the fallacy of algorithmic neutrality. Solutions for climate and food security should look toward regenerative and ancestral technologies rather than the extractive logic of Big Tech. Future development must be subordinated to socio-environmental justice, emphasizing cooperation and mutual care over the dystopian model of patriarchal, capital-driven domination.

  Global Voices · Apr 23, 2026

  

• India’s AI Adoption: Balancing Innovation and Human Rights

  India’s AI governance currently relies on non-binding "soft law" guidelines that prioritize development over mandatory transparency or human rights impact assessments. State-led AI deployment, particularly in surveillance and welfare, is increasingly impacting marginalized communities and civic space. Experts and human rights organizations argue for legally binding legislation to address bias, accountability, and the risks of digital authoritarianism. Current State of AI Governance India lacks a dedicated, binding AI statute, utilizing existing frameworks like the IT Act and the 2025 Data Protection Rules. The "Artificial Intelligence (Ethics and Accountability) Bill, 2025" has been proposed but not enacted; it aims to introduce mandatory ethical reviews and grievance mechanisms. MeitY’s current "IndiaAI Mission" emphasizes "Safe & Trusted AI" through voluntary projects, which critics argue lack the teeth of effective, enforceable oversight. Pervasive Surveillance Infrastructure During the February 2026 India AI Impact Summit, Delhi was transformed into a "digital fortress" using 4,000+ AI-enabled cameras and 32 control rooms for real-time monitoring. Project Panoptic, led by the Internet Freedom Foundation (IFF), has tracked over 120 government contracts involving facial recognition technology across India. The "DigiYatra" aviation app has faced criticism for weak data security and opacity, as the Digi Yatra Foundation is largely controlled by private entities, exempting it from the Right to Information Act. Impact on Vulnerable Populations AI-driven facial recognition in the "Poshan Tracker" welfare app has led to the denial of food benefits for millions of women and children, as the system fails to account for appearance changes due to aging, pregnancy, or illness. Biased algorithms disproportionately affect those with darker skin tones and those relying on low-end hardware, exacerbating existing social inequalities. Similar surveillance technologies are being deployed in schools for exam proctoring, raising concerns over constant monitoring and privacy intrusion. Calls for Reform Human rights bodies like Amnesty International and the Freedom Online Coalition advocate for mandatory human rights due diligence and the banning of invasive biometric surveillance. Advocates call for specific, enforceable legislation that mandates public disclosure of AI data training sets, error rates, and accountability structures. There is a growing consensus that human rights, rather than being an afterthought, must be integrated into the fundamental design of AI systems.

  Global Voices · Apr 23, 2026

  

• For AI to Work for Us, It Will Have to Stop Pretending to Be Us

  AI is not neutral; it actively reproduces existing patriarchal and colonial power structures and biases present in its training data. Tech companies prioritize business models and profit over human rights, frequently reducing human lives to data points and disposable inputs. AI cannot replicate human empathy, lived experience, or moral responsibility; attempts to simulate these qualities devalue human connection. A human rights-based approach requires moving beyond the myth of technological inevitability to hold power accountable for the societal impacts of their products. The Myth of Neutrality AI models scrape vast amounts of data that contain historical exclusions, racism, and inequality, which the system then encodes and amplifies. Corporate incentives drive the deployment of advanced AI without adequate understanding of the consequences, impacting land, resources, and ecological health. In military or policing contexts, the algorithmic reduction of human beings to targets or success rates facilitates the justification of harm and dehumanization. Why AI Cannot Replace Humans AI operations are based on statistical probability and pattern recognition, which lacks the context, care, and accountability inherent in human relationships. Treating machine-generated outputs as equivalent to human interaction erodes the value of community, empathy, and collective resistance. The narrative that AI replaces humans is a dangerous framing that enables the systemic disposability of people in sectors like welfare, health care, and hiring. Demanding Accountability The burden of mitigating AI harm should not fall on those with the least power; accountability must rest with the entities building and deploying the technology. We must interrogate the design of these systems at the point of commercialization, asking who benefits and what values are being prioritized. Future development should be centered on the needs of living beings rather than optimizing systems for corporate growth and mechanical efficiency.

  Global Voices · Apr 22, 2026

  

• AI Warfare Triggers Putin as Kremlin Dismantles Russian Internet

  The Russian government has intensified efforts to restrict internet access through unprecedented mobile shutdowns, a crackdown on VPNs, and the implementation of "white lists." Anxiety in the Kremlin, allegedly triggered by AI-powered military operations against Iran, has led to a push for total digital control within Russia. Major Russian digital platforms, including Yandex and VK, are now being forced to block VPN users and spy on their customers to report data to authorities. AI Warfare and Geopolitical Paranoia The US operation "Epic Fury" against Iran, utilizing Project Maven AI, successfully targeted multiple government and military officials using accelerated AI-driven decision-making. Political theory professor Elke Schwarz warned that AI-led strikes lack human oversight and suffer from low reliability (25–50%), creating dangerous geopolitical precedents. Reports suggest that Putin’s recent paranoia and the severe mobile internet shutdowns in Moscow—lasting weeks—are direct responses to fears that similar AI-targeting technology could be turned against Russian leadership. The Dismantling of the Russian Internet Internet shutdowns are reportedly being directed by the FSB’s Scientific and Technical Service (STS), which maps specific districts for forced blackouts. Despite a state-mandated block on Telegram, approximately 65 million Russians continue to use the messenger daily. New regulations mandate that Russian apps monitor user devices for VPN tools, effectively turning domestic platforms into state-mandated spyware. AI Development and Military Sovereignty Putin has shifted his rhetoric toward the militarization of AI, demanding domestic LLM development for national defense. Russian developers are actively leveraging open-source foreign models like LLaMA and Mistral to build autonomous drones and guidance systems, bypassing the need for domestic innovation from scratch. Russia’s history of AI exports, such as the FindFace facial recognition system used by Iran to suppress protests, highlights the dual-use nature of the country's technological capabilities.

  Global Voices · Apr 18, 2026

  

• Indonesia and the Politics of Platform Governance

  Indonesia is intensifying efforts to regulate global social media platforms, citing the need to combat misinformation and disinformation under the Electronic Information and Transactions Law (UU ITE). Government interventions include mandatory registration as Electronic System Operators (PSE), content removal orders via the SAMAN system, and pressure to provide user data during protests. Experts warn that vague regulatory language regarding "harmful content" risks suppressing dissent and violating rights to information, echoing historical state control over traditional media. Regional examples from Cambodia and Myanmar highlight that platform governance is a complex negotiation between government demands, corporate algorithms, and geopolitical pressures. Regulatory Framework and Enforcement In March 2026, the Minister of Communication and Digital Affairs, Meutya Hafid, inspected Meta’s office to demand greater compliance and algorithmic transparency. Under the 2020 PSE regulations, firms failing to register risk fines or blocking by ISPs; platforms like Wikimedia have faced access restrictions for non-compliance. The SAMAN system enables the government to force platforms to remove content within 4–24 hours, with potential fines reaching 500 million IDR (over USD 29,000) per violation. Concerns persist regarding "mission creep," where authorities allegedly request data on users monetizing livestreams or target content critical of government actions, such as reporting on 1998 riots or environmental issues. Historical Context and Media Control Current platform regulation mirrors the pre-Reformasi era, where state oversight and licensing limited journalistic autonomy. Modern media ownership in Indonesia remains concentrated among a small group of conglomerates, creating a "media oligarchy" that influences political coverage during electoral cycles, as seen in the support patterns of networks like Metro TV and TVOne. Comparative Regional Dynamics Cambodia: Prime Minister Hun Sen used threats against political opponents on Facebook; the Meta Oversight Board recommended a suspension, leading to a standoff where Meta complied with the video removal but avoided a full account suspension to maintain presence. Myanmar: Facebook was criticized for failing to moderate hate speech against the Rohingya; despite legal barriers like Section 230 in the US, the resulting reputational and regulatory pressure forced the company to reform its content moderation policies. Platform-government relationships are rarely unidirectional; they involve a balance of legal, regulatory, and reputational risks for corporations against the enforcement power of sovereign states.

  Global Voices · Apr 17, 2026