Lorem Ipsum has been the industry’s standard dummy text ever since the 1500s.
Introduction
When a self-driving car recognizes a stop sign or when facial recognition software identifies a person in a crowd, the technology appears seamlessly automated. Yet behind these AI marvels lies an invisible workforce of human laborers who painstakingly labeled millions of images, videos, and data points to train these systems. In India and the Philippines—two critical hubs for AI outsourcing—these “ghost workers” toil in precarious conditions for wages as low as $0.01 per task, powering a multi-billion-dollar industry that rarely acknowledges their existence.
This digital labor exploitation represents a new form of technological colonialism, where Global North companies extract value from Global South workers while offering minimal compensation and protection. As the AI industry races toward a projected $17.1 billion market by 2030, the human cost of this growth remains deliberately obscured. This analysis exposes the systemic exploitation underlying AI development and argues for urgent regulatory and ethical reforms to ensure that the benefits of artificial intelligence are shared more equitably with those who make it possible.
The Architecture of AI Exploitation
The artificial intelligence industry’s dependence on human labor is both extensive and intentionally hidden. Despite AI’s promise of automation, the technology requires massive amounts of human-labeled training data. Workers must identify objects in images, transcribe audio, moderate content, and perform countless other tasks that teach algorithms to recognize patterns. This work is then outsourced to countries like India and the Philippines, where companies can exploit lower wages and weaker labor protections.
Nearly half of all global online freelance work now occurs in India and the Philippines, with their combined 360 million-person labor force making them prime targets for outsourcing.¹ Companies like Scale AI, iMerit, and SamaSource have built business models around this geographic arbitrage, paying workers a fraction of what similar labor would cost in developed countries while charging premium rates to AI companies.²
The scale of this hidden economy is staggering. The global data annotation market alone employs millions of workers, yet most operate without traditional employment protections, benefits, or job security. They exist in a regulatory gray area where they are classified as independent contractors rather than employees, absolving companies of most labor law obligations.
India: The Data Labeling Factory
Economic Exploitation
In India, data labeling has become a poverty wage occupation despite requiring significant skill and attention to detail. Workers typically earn between ₹15,000 to ₹20,000 per month ($180-$240), with some sources reporting wages as low as ₹15,000 monthly—figures that have remained stagnant since 2019 despite the industry’s explosive growth.³ While some specialized roles may command higher wages, potentially reaching ₹42,458 monthly ($510), these represent outliers rather than typical compensation.⁴
These wages pale in comparison to similar work in developed countries, where data labelers earn $50,000-$67,000 annually—roughly ten times more for identical tasks.⁵ The wage disparity cannot be explained by cost of living differences alone; it reflects a deliberate strategy to maximize profit margins by exploiting global economic inequalities.
Working Conditions and Health Impacts
Indian data labelers work in overcrowded, often dusty environments that lack basic workplace amenities. Most workers receive no healthcare benefits, sick leave, or worker’s compensation despite performing repetitive tasks that can cause physical strain and mental stress.⁶ The nature of the work itself compounds these problems—workers frequently encounter disturbing content including graphic violence, hate speech, and explicit material while moderating social media posts or training content recognition systems.
The psychological toll is severe but rarely acknowledged. Workers report anxiety, depression, and trauma from constant exposure to disturbing content, yet they receive no mental health support or counseling services. Performance evaluations are opaque and arbitrary, leaving workers vulnerable to sudden termination without recourse or explanation.
The Ghost Worker Phenomenon
Indian AI workers describe themselves as “ghost workers”—essential to the technology but invisible to end users and often to the companies that profit from their labor.⁷ They train algorithms for self-driving cars, facial recognition systems, and recommendation engines that generate billions in revenue, yet their contributions remain unacknowledged.
Language barriers compound their vulnerability. Instructions are typically provided in English, which many workers may not fully understand, leading to confusion, errors, and termination. This linguistic colonialism ensures that workers remain at a disadvantage in systems designed to extract their labor while minimizing their agency.
The Philippines: Digital Sweatshops and AI Deception
Extreme Wage Suppression
The Philippines represents perhaps the most egregious example of AI labor exploitation. Filipino data labelers earn below the country’s $6-$10 daily minimum wage, with many tasks on platforms like Scale AI’s Remotasks paying as little as $0.01 per completed item.⁸ Payments are frequently delayed, reduced, or withheld entirely, creating a climate of financial uncertainty that keeps workers desperate for any available work.
This wage suppression occurs within a $20 billion “ghost work” industry where workers lack job security, benefits, or legal protections.⁹ The 1.84 million business process outsourcing (BPO) workers in the Philippines face an existential threat, with an estimated 300,000 jobs at risk of elimination within five years as companies increasingly automate or relocate operations.¹⁰
High-Pressure Surveillance and Control
Filipino workers operate under intense surveillance and productivity monitoring that would be illegal in many developed countries. Their computer screens are monitored, keystrokes tracked, and output measured in real-time, creating a panopticon of digital control that maximizes extraction while minimizing worker autonomy.
The psychological pressure is immense. Workers report constant stress, burnout, and anxiety from the knowledge that their every action is being measured and evaluated. Unlike traditional employment relationships, they have no collective bargaining power, workplace representation, or mechanisms to address grievances.
The Faux AI Scandal
A 2025 scandal exposed the extent of deception within the AI industry when a U.S. startup was revealed to be using Filipino workers to simulate AI behavior for customer service and data analysis.¹¹ Workers were instructed to respond as if they were artificial intelligence, performing complex cognitive tasks while being paid a fraction of what genuine AI development would cost.
This “faux AI” practice represents the logical extreme of AI labor exploitation—workers not only provide the training data for artificial intelligence but are forced to pretend to be AI systems themselves. The scandal sparked global outrage but also revealed how normalized such practices have become within the industry.
Workers like Renso Bajala, who was fired for speaking publicly about these conditions, exemplify the “code of silence” that protects exploitative practices.¹² Their testimonies reveal systematic wage theft, psychological manipulation, and deliberate dehumanization designed to maximize profit extraction.
Regulatory Failures and Emerging Responses
India’s Policy Vacuum
India currently lacks specific regulations governing AI labor practices, despite hosting millions of such workers. While the Ministry of Electronics and Information Technology issued advisories in 2024 requiring platforms to seek permission before deploying AI models, these focus on bias and discrimination rather than worker protection.¹³
Proposed amendments to Indian labor laws acknowledge AI’s potential impact on employment and privacy, but concrete protections remain theoretical.¹⁴ An EY report estimates that AI could affect 38 million Indian employees by 2030, yet policymakers have done little to prepare for this transition or protect vulnerable workers.¹⁵
The Philippines’ Emerging Framework
The Philippines has made more progress toward addressing AI labor exploitation through the proposed Artificial Intelligence Regulation Act. Advocacy coalitions like Code AI are pushing for stronger labor protections and free speech safeguards within the legislation.¹⁶
The Department of Labor and Employment (DOLE) and Department of Information and Communications Technology (DICT) are developing ethical outsourcing guidelines that could provide some worker protections.¹⁷ However, the influence of powerful BPO industry lobbying groups threatens to weaken these measures before implementation.
The Global Context: Digital Colonialism
The exploitation of AI workers in India and the Philippines represents a new form of colonialism adapted for the digital age. Just as historical colonialism extracted raw materials from the Global South to fuel industrial development in the Global North, digital colonialism extracts human intelligence and labor to fuel AI development in wealthy countries.
This system perpetuates global inequality by concentrating AI profits in Silicon Valley and other tech hubs while distributing only subsistence wages to the workers who make the technology possible. The cognitive labor that creates AI’s value is systematically devalued when performed by Global South workers, even as identical work commands premium compensation in developed countries.
Major tech companies benefit from this arrangement through multiple layers of abstraction. They contract with AI companies like Scale AI, which in turn subcontract to local firms in India and the Philippines, creating plausible deniability about working conditions and wages. This supply chain obfuscation allows companies to claim ethical practices while benefiting from exploitative labor.
Consequences and Implications
Impact on AI Quality and Safety
The exploitation of AI workers doesn’t just harm individuals—it threatens the quality and safety of AI systems themselves. Overworked, underpaid, and psychologically stressed workers are more likely to make errors in data labeling, potentially introducing biases or mistakes that propagate through AI systems. When workers lack proper training, support, or compensation, the accuracy and reliability of AI outputs suffer.
The high turnover rates caused by poor working conditions mean that institutional knowledge is constantly lost, requiring new workers to learn complex tasks without adequate training or support. This cycle of exploitation and turnover undermines the very foundation of AI development.
Perpetuating Global Inequality
AI labor exploitation reinforces existing patterns of global inequality by ensuring that the economic benefits of technological advancement flow primarily to wealthy countries and corporations. While AI promises to transform economies and societies, its current development model ensures that this transformation benefits primarily those who already possess capital and power.
The workers who create AI’s value are excluded from sharing in its benefits, trapped in precarious employment relationships that offer no path to economic advancement or skill development. This pattern risks creating a permanent underclass of digital laborers whose intelligence and effort enrich others while leaving them in poverty.
Toward Ethical AI Development
Immediate Reforms
Addressing AI labor exploitation requires immediate action across multiple domains. Companies must implement transparent wage standards that ensure AI workers receive compensation comparable to similar work in developed markets, adjusted for local cost of living. This means ending the practice of paying workers $0.01 per task or below minimum wage for skilled cognitive labor.
Worker classification must also be reformed. The current practice of classifying AI workers as independent contractors while exercising employee-level control over their work violates both the spirit and often the letter of labor law. Workers deserve employment protections, benefits, and collective bargaining rights.
Regulatory Solutions
Governments in both source and destination countries must act to regulate AI labor practices. India and the Philippines need comprehensive AI labor protection laws that establish wage floors, working condition standards, and worker rights. These laws should include provisions for mental health support, particularly for workers exposed to disturbing content.
Developed countries must also take responsibility by regulating companies that profit from exploitative AI labor. Corporate accountability laws should require companies to audit their AI supply chains and ensure ethical labor practices throughout their operations.
Industry Transformation
The AI industry must acknowledge that current labor practices are unsustainable and unethical. This means moving beyond corporate social responsibility rhetoric to implement structural changes in how AI development is funded, organized, and executed.
Worker ownership models, profit-sharing arrangements, and cooperative structures could ensure that those who create AI’s value share in its benefits. Companies could also invest in worker development, providing training and career advancement opportunities that allow AI workers to transition into higher-value roles within the industry.
Global Cooperation
International cooperation is essential to address the global nature of AI labor exploitation. This could include international labor standards specifically designed for digital work, technology transfer programs that build AI capabilities in developing countries, and trade agreements that include strong labor protections.
The goal should be to transform AI development from an extractive industry into one that genuinely benefits all participants, creating sustainable economic opportunities while advancing technological capabilities.
Conclusion
The artificial intelligence revolution is built on the invisible labor of millions of workers in India, the Philippines, and other developing countries. These “ghost workers” train the algorithms that power self-driving cars, facial recognition systems, and countless other AI applications, yet they remain trapped in exploitative working conditions that offer minimal compensation for their essential contributions.
This exploitation represents more than individual corporate malfeasance—it reflects systemic patterns of digital colonialism that concentrate AI’s benefits in wealthy countries while distributing only subsistence wages to those who make the technology possible. As AI becomes increasingly central to global economic and social systems, these inequities will only deepen unless decisive action is taken.
The path forward requires acknowledging that AI is not a purely technological phenomenon but a socio-economic system that embeds particular values and power relationships. If we want AI to serve humanity broadly rather than just a privileged few, we must ensure that its development is based on principles of equity, dignity, and shared prosperity rather than extraction and exploitation.
The workers who train our AI systems deserve more than poverty wages and precarious employment. They deserve recognition as essential contributors to one of the most important technological developments in human history, with compensation and working conditions that reflect the value of their contributions. Until we address this fundamental injustice, the promise of artificial intelligence will remain unfulfilled, built on a foundation of exploitation that undermines its potential to benefit humanity as a whole.
The time for reform is now, before these exploitative patterns become even more entrenched in our technological infrastructure. The future of AI—and the workers who make it possible—depends on the choices we make today.
Notes
¹ Washington Post, “Scale AI’s Remotasks workers in the Philippines”
² The Conversation, “AI is a multi-billion dollar industry”
³ Glassdoor, “Data Labeling Salary in India”; Times of India, “How artificial intelligence is creating jobs in India”
⁴ ExtraPe, “Earn Money with Data Labeling”
⁵ Glassdoor, “Data Labeling Salary in India”
⁶ The Conversation, “AI is a multi-billion dollar industry”
⁷ Noema Magazine, “The Exploited Labor Behind Artificial Intelligence”
⁸ Washington Post, “Scale AI’s Remotasks workers in the Philippines”
⁹ Filipinos in the 6ix, “Filipino Workers Posed as AI”
¹⁰ Rest of World, “Filipino tech workers demand protections in AI bill”
¹¹ Filipinos in the 6ix, “Filipino Workers Posed as AI”
¹² Rest of World, “Filipino tech workers demand protections in AI bill”
¹³ India Briefing, “Regulation of AI and Large Language Models in India”
¹⁴ International Journal of Legal Science, “Rethinking Indian Labour Law and Policy”
¹⁵ EY Report, “How much productivity can GenAI unlock in India”
¹⁶ Rest of World, “Filipino tech workers demand protections in AI bill”
¹⁷ Filipinos in the 6ix, “Filipino Workers Posed as AI”