The AI Agent Is Learning Workers' Rights

Editorial Summary: The Indian Express argues that AI agents — autonomous systems that plan and execute workplace decisions without human intervention — are fundamentally different from conventional tools, and treating them as such leaves gig and platform workers without recourse when algorithms make discriminatory, opaque, or arbitrary employment decisions. India’s Labour Codes and even Rajasthan’s Platform Workers Act 2023 do not address algorithmic accountability. The editorial calls for amendments to the Code on Social Security to mandate human oversight for high-stakes AI decisions, a right to explanation for algorithmic employment choices, and India’s active participation in the ILO’s global standards dialogue on algorithmic management.

The Agent, Not the Tool

In January 2026, a global logistics company deployed AI agents across its warehouse network — autonomous systems that assigned tasks, monitored worker pace, flagged performance deviations, and issued warnings, all without a human supervisor reviewing individual decisions. Within three months, worker grievance filings at several facilities had doubled. The complaints were consistent: workers did not know why they were being evaluated poorly, could not appeal to a human supervisor, and could not identify who was responsible for the decisions affecting their employment.

The episode illustrates a governance gap that is rapidly widening. AI agents are not chatbots. They do not simply respond to queries. They plan multi-step tasks, infer goals from context, interact with third-party services, and make consequential decisions autonomously. The distinction between AI as a tool and AI as an autonomous actor is not philosophical — it is legally and ethically load-bearing.

What Is an AI Agent?

A conventional software system executes pre-defined instructions: if X, then Y. An AI agent operates differently. Using large language models (LLMs) or other foundation models as a reasoning engine, an agent can: interpret a high-level goal stated in natural language, break it into sub-tasks, call external tools and APIs, and complete complex workflows without step-by-step human instruction.

Examples of agentic AI in workplaces:

The common thread: decisions that previously required a human supervisor’s judgment are now made by systems that learn from data, adapt to context, and operate without a human in the loop for individual decisions.

Algorithmic Management: India’s Gig Economy

India’s gig economy is among the world’s largest and fastest-growing. The NITI Aayog 2022 report estimated 7.7 million gig workers employed through digital platforms; projections suggest this could reach 23.5 million by 2029-30. These workers are governed, in the most direct sense, by algorithms:

• Uber and Ola: Driver rating systems, surge pricing, ride allocation, and — crucially — platform deactivation are all AI-driven. A driver whose rating falls below a threshold may be deactivated without a human reviewing the individual case history.
• Swiggy and Zomato: Delivery agent performance scoring, shift allocation, and incentive payments are governed by algorithmic systems. Workers report that opaque metrics govern whether they receive premium delivery slots.
• Urban Company: Household service professionals are rated, ranked, and promoted or demoted by AI systems that aggregate customer feedback.

In every case, the workers governed by these systems lack: transparency about the criteria used to evaluate them, an ability to challenge algorithmic decisions, and a clear identification of a human decision-maker responsible for outcomes.

The Accountability Gap

The fundamental legal problem with agentic AI in employment is the accountability gap. Labour law — from India’s Industrial Disputes Act 1947 (now subsumed in the Industrial Relations Code 2020) to international conventions — assumes a human employer who can be held responsible for employment decisions. When an AI agent makes a discriminatory hiring decision, denies a gig worker a shift, or initiates a platform deactivation:

• Who is liable? The AI? It has no legal personality. The developer who built it? The deployer (the platform company) who configured it? The manager who set the performance targets that the algorithm optimised for?
• What is the standard? Under the Equal Remuneration Act and the Maternity Benefit Act (now Labour Codes), intent is often an element of discrimination claims. An AI agent may discriminate statistically — producing disparate outcomes for women, minorities, or disabled workers — without any individual human having intended the outcome.
• What is the remedy? Without a human decision-maker, the grievance redressal architecture of the Industrial Tribunal system has no respondent to summon.

This is not a theoretical concern. Research in the United States and European Union has documented AI hiring systems that penalise CVs containing the word “women’s” (as in women’s college), AI scheduling systems that systematically allocate fewer hours to workers with caregiving obligations, and AI performance systems that rate older workers lower due to training data biases.

The EU AI Act: A Reference Point

The EU AI Act, adopted in 2024 and entering full application in phases, directly addresses AI in employment. It classifies the following as high-risk AI systems:

• AI used in recruitment and selection (CV screening, interview assessment, shortlisting)
• AI used to make or assist decisions on work organisation, task allocation, and work performance evaluation
• AI used for real-time biometric monitoring of workers
• AI used for access to self-employment (i.e., gig platform onboarding and deactivation)

High-risk AI systems under the EU AI Act must: undergo conformity assessment before deployment, maintain human oversight provisions, provide transparency to affected workers, enable meaningful human review of consequential decisions, and maintain audit logs for regulatory inspection.

The EU framework does not ban AI in employment. It mandates accountability infrastructure. India has no equivalent provision.

India’s Labour Framework: What Exists and What Is Missing

India’s four Labour Codes (2019-20) consolidate 29 central labour laws:

The Code on Social Security 2020 is the most significant advance: it defines gig workers and platform workers for the first time in Indian central legislation and provides for their inclusion in social security schemes. Rajasthan’s Platform Workers Act 2023 operationalised this at the state level — creating a registration mechanism and a welfare fund.

Neither instrument addresses algorithmic accountability. The Codes do not: require platforms to disclose the criteria by which AI systems evaluate workers, mandate human review of AI-driven deactivations or performance penalties, assign liability to platform operators for discriminatory algorithmic outcomes, or create a right for workers to challenge algorithmic decisions.

ILO and the Global Standards Gap

The International Labour Organization’s Global Commission on the Future of Work (2019) identified “algorithmic management” as one of three structural changes reshaping work — alongside automation and the gig economy. The Commission recommended: portability of social protection, investment in lifelong learning, and a human-in-command principle for AI in employment.

The ILO has since engaged in technical standard-setting on digital labour platforms, producing a non-binding code of conduct for platform operators. The ILO-EU dialogue on algorithmic management has produced principles: transparency, contestability, human oversight, and non-discrimination. But no binding ILO Convention on algorithmic management exists.

India is a member of ILO’s governing body. Participation in the standards dialogue — and advocacy for a binding Convention — would be consistent with India’s status as the world’s largest gig economy by workforce size.

UPSC Mains Analysis

GS Paper 2 — Governance, Social Justice; GS Paper 3 — Science and Technology; GS Paper 4 — Ethics and accountability

Key arguments:

• AI agents are qualitatively different from tools — they make autonomous employment decisions without a human in the loop, creating an accountability gap.
• India has 7.7 million gig workers (NITI Aayog 2022) governed by algorithmic management systems on platforms like Uber, Swiggy, Ola, and Urban Company.
• Labour Codes 2019-20 and Rajasthan Platform Workers Act 2023 are advances; neither addresses algorithmic accountability.
• EU AI Act 2024 classifies AI in employment as high-risk, mandating human oversight, transparency, and conformity assessment — India has no equivalent.
• ILO Global Commission on Future of Work 2019: human-in-command principle for AI employment decisions; no binding ILO Convention yet.
• Code on Social Security 2020: gig/platform worker definition and social security inclusion — but no algorithmic liability provisions.

Counterarguments:

• Excessive regulation may deter AI adoption in employment, slowing productivity gains that benefit workers through higher wages and reduced drudgery.
• ILO research shows AI augments more tasks than it eliminates in aggregate — displacement narrative may be overstated.
• Compliance costs for algorithmic accountability frameworks may disproportionately burden smaller gig platforms relative to large incumbents.

Keywords: AI agents, algorithmic management, gig economy, 7.7 million gig workers NITI Aayog 2022, Rajasthan Platform Workers Act 2023, Code on Social Security 2020, Labour Codes 2019-20, EU AI Act 2024 high-risk AI, ILO Global Commission Future of Work 2019, platform deactivation, accountability gap, Uber, Swiggy, large language models (LLM), agentic AI.

Editorial Insight

The Indian Express’s argument is that the arrival of AI agents in the workplace is not an extension of automation — it is a transfer of managerial authority from humans to systems that have no legal identity, no accountability to labour tribunals, and no obligation to explain themselves to the workers they govern. India’s gig workers bear the front-line cost of this governance gap. The Code on Social Security 2020 was a landmark; an algorithmic accountability amendment would be the next one. India cannot wait for the EU to set the global standard on a problem that India’s 7.7 million gig workers are already living.

Sources: Indian Express, NITI Aayog, ILO, PIB