US Companies Accused of Exploiting Indian Workers in the Shadowy World of AI Data Labor

The rapid rise of artificial intelligence has transformed industries, powering everything from chatbots and recommendation algorithms to self-driving cars and content moderation systems. Behind the glossy promises of seamless automation lies a less visible reality: a massive global workforce performing tedious, often distressing tasks to train these AI models. In India, thousands of workers are at the forefront of this “data economy,” labeling images, annotating text, moderating harmful content, and even capturing real-world footage via head-mounted cameras. Critics argue that US tech giants are exploiting this labor pool through low wages, precarious conditions, and opaque subcontracting chains, reaping billions while Indian workers bear the human cost.

This phenomenon highlights a broader pattern in the AI supply chain. Major American companies outsource foundational work to countries in the Global South, including India, where labor is abundant, skilled, and cost-effective. As the AI industry hurtles toward a projected multi-hundred-billion-dollar valuation, questions about ethics, fairness, and sustainability grow louder.

The Mechanics of AI Data Work in India

Data labeling, or annotation, involves humans tagging raw data so machine learning models can “learn” patterns. For instance, workers might draw bounding boxes around objects in images for computer vision systems, categorize toxic text for safety filters, or rank AI-generated responses for quality. In India, this work spans urban BPO centers, rural “cloud farms,” and remote gig platforms.

One striking example involves factory or household workers wearing headcams to record daily activities. Companies collect this first-person footage to train robotics AI on human behaviors — essentially paying Indians modest sums to help machines potentially replace similar roles in the future. Reports indicate payments ranging from ₹1,500 to ₹5,000 per hour for such specialized data capture, though average rates for standard labeling are far lower.

According to Nasscom, as early as 2021, around 70,000 people in India worked in data annotation, generating a market worth about $250 million, with 60% of revenues from the US. This ecosystem includes local firms and international players like Scale AI (valued highly and backed by prominent investors), Appen, iMerit, and others that act as intermediaries for US clients such as OpenAI, Meta, Google, Microsoft, and Amazon.

Workers often operate under non-disclosure agreements (NDAs), making their contributions invisible. Tasks arrive via digital platforms where algorithms monitor speed, accuracy, and compliance. Deviations can lead to task rejection or account suspension, leaving workers without recourse.

Low Wages and Precarious Employment

A core allegation of exploitation centers on compensation. While US-based data workers might earn significantly more, Indian counterparts frequently receive wages that translate to low hourly rates — sometimes equivalent to less than $2 for demanding tasks, though figures vary by project and platform. Many juggle microtasks on sites reminiscent of Amazon Mechanical Turk, where pay depends on volume and approval rates. One Indian worker with an engineering background reported earning $10–30 daily after completing numerous tasks, often working nights to align with US client hours.

Contracts are typically short-term or gig-based, offering little job security, health benefits, or paid leave. Algorithmic management adds pressure: workers face constant surveillance, with systems flagging idle time or errors. This “fissured” employment model — where Big Tech distances itself through layers of subcontractors — allows companies to scale rapidly while minimizing direct liabilities.

In rural and semi-urban areas, these jobs are pitched as opportunities for graduates, women, and underserved communities. Platforms like Karya promote fairer wages and local empowerment. However, broader industry reports suggest many workers remain trapped in low-skill cycles with thin career progression, risking burnout as AI demand surges.

The Psychological and Health Toll

Beyond finances, the nature of the work inflicts hidden damage. Content moderators and labelers routinely review graphic violence, hate speech, child exploitation material, and explicit content to train AI guardrails. Indian female workers have described feeling emotionally “blank” after hours of exposure, with limited mental health support.

Similar issues plague workers globally, from Kenya to the Philippines. Studies link prolonged exposure to traumatic material with anxiety, depression, and other issues. In India, where mental health resources can be stretched, this burden is particularly concerning. NDAs and the gig structure discourage open discussion or collective bargaining.

Additionally, some workers train AI systems that could automate their own jobs or similar ones, creating an existential irony. By feeding data into models for robotics or process automation, they accelerate a future where human roles diminish — all for modest, immediate pay.

Global Context and US Companies’ Role

India is not alone. Parallel exploitation stories emerge from Kenya (where OpenAI contractors earned under $2/hour for toxic content labeling), the Philippines, Venezuela, and Colombia. US firms dominate as clients because the AI boom requires enormous labeled datasets — far beyond what domestic labor could supply affordably.

Intermediaries like Scale AI have faced scrutiny, including US Department of Labor probes into wage practices. Open letters from African data workers to US officials accuse Big Tech of undermining labor laws and enabling “modern-day slavery” conditions. In India, the dynamic mirrors these patterns, amplified by the country’s massive talent pool and competitive pressures.

Proponents argue outsourcing creates jobs and transfers skills. India’s improving connectivity and English proficiency make it attractive. Yet critics, including researchers and ethicists, point to colonial-era extractive dynamics: wealth flows northward while risks stay local. Transparency is minimal; consumers using AI tools rarely know about the human backbone.

Growing Pushback and Potential Solutions

Resistance is building. In various regions, data workers form associations and unions demanding living wages, psychological support, and fair contracts. The Global Trade Union Alliance of Content Moderators and similar groups advocate across borders. In India, calls for stronger regulations on gig platforms and BPOs could help.

Ethical alternatives exist. Some platforms prioritize fair pay and upskilling. Policymakers could incentivize these models, enforce minimum standards, and require Big Tech to disclose labor practices in supply chains. US lawmakers have sent letters urging transparency and better conditions for “ghost workers.”

For workers, diversification into higher-value AI roles — such as domain-specific annotation in medicine or law — offers pathways, but requires investment in training.

Implications for the Future of AI and Work

The exploitation narrative underscores a fundamental tension: AI’s success depends on undervalued human intelligence. Without addressing it, the industry risks backlash, talent shortages, and ethical scandals that could slow innovation.

For India, a leader in IT and outsourcing, this presents both opportunity and challenge. As AI adoption accelerates domestically and globally, ensuring workers are partners rather than disposable inputs is vital. Broader economic policies supporting skill upgrades, social protections for gig workers, and data sovereignty could mitigate downsides.

Ultimately, US companies’ reliance on Indian data labor reveals the limits of “artificial” intelligence. True progress requires valuing the humans who make machines smarter. As awareness grows through investigations and worker testimonies, pressure for reform will likely intensify. The question remains whether the AI revolution will uplift its foundational workforce or perpetuate inequality in the global digital economy.

About The Author