Lessons from Chile and Uruguay for making the new ILO platform-work standard enforceable
By María Eugenia Rodríguez Florez, International Consultant and Independent Scholar
This blog is produced as a part of the Reshaping Work Fellowship Programme. The opinions and views expressed in this publication are those of the author. They do not purport to reflect the opinions or views of Reshaping Work or organisations that have supported the programme.
The adoption of the Decent Work in the Platform Economy Convention, 2026, at the 114th Session of the International Labour Conference marks an extraordinary achievement after more than a decade of advocacy, litigation, organizing, and research showing that platform workers should not be excluded from labour protections simply because work is mediated by an app, classified as independent, or organized through automated systems. Its most innovative contribution lies in the provisions on algorithmic management. By requiring platforms to disclose not only the use of automated systems, but also their impact on working conditions and access to work, and by linking this disclosure to responsible use, written explanations, review of adverse decisions, and appropriate human involvement, the Convention establishes a new international benchmark: algorithmic management is no longer only a technical or data-governance issue, but a labour-rights issue.
This achievement also creates a strategic opportunity to build countervailing power. If implemented effectively, the Convention can strengthen the ability of workers, unions, and regulators to challenge the informational asymmetry at the center of platform work: platforms control the data, the rules of allocation, the rating systems, the pay architecture, and the mechanisms of suspension or deactivation. In this sense, the Convention connects directly with Pulignano V. (2024)’s concept of algorithm governance, which moves beyond narrow data-privacy concerns by showing how algorithmic management is embedded in contractual arrangements, labour-market regulation, and collective representation. Read together, the Convention’s provisions on collective rights, the legal conditions under which platform work is organized, the impact of automated systems, written explanations, human review, remedies, and enforcement suggest that algorithmic management should be governed as part of labour regulation, not treated as an autonomous technical system. The central tension, therefore, is whether AI governance in platform work will remain a formal language of transparency and review, or become an enforceable framework through which workers, unions, and regulators can contest algorithmic control.
The question for Latin America is whether this new international standard can be translated into domestic platform-work laws that are not only formally protective, but also enforceable. The region is not a single regulatory model. It contains different legal traditions, inspection systems, levels of informality, and forms of worker organization. Precisely for that reason, Latin America provides a useful setting to examine how the Convention’s benchmark may interact with domestic institutional capacity, fragmented employment relations, and significant asymmetries of information vis-à-vis transnational platforms.
This Latin American question is not hypothetical. The region already includes countries that have attempted to regulate platform work, and Chile and Uruguay offer two useful illustrations. Although Chile’s and Uruguay’s laws predate the 2026 Convention, they provide useful evidence on the domestic enforcement challenges that the new international standard is likely to encounter. Rosenbaum Carli (2026) frames the comparison as one between “a country with a formal regulatory framework” and one where “protections were primarily shaped by the judiciary,” focusing on “regulatory trajectories and enforcement mechanisms.” These cases do not represent the whole region, but they help examine whether domestic laws can translate the Convention’s concern with automated systems, access to work, income, sanctions, and human review into enforceable labour rights.
Chile’s Law 21.431, enacted in March 2022, addresses algorithmic management through several dispersed provisions. It requires dependent-worker contracts to include the “impact of user ratings” on the worker. It creates a complaint channel for payments, work records, task assignment, and client evaluations, which must “always be attended by a person” if the worker requests. It also requires the criteria used to establish contact and coordination with users to be “transparent and objective.” More directly, platforms must allow competent authorities, when required, access to the “programming of the algorithm” and to “complete and sufficient explanations” of how it makes decisions. The law also prohibits discrimination through automated decision-making mechanisms and requires equality and non-discrimination in the “implementation of algorithms,” especially in “task allocation, bonuses, incentives, and remuneration.”
Uruguay’s Law 20.396, approved in February 2025, is more explicit in its terminology. It regulates “transparency of algorithms and monitoring systems” and requires information on automated monitoring and decision-making systems affecting assignments, income, working time, contractual status, and account restriction, suspension, or cancellation. At the same time, it expressly limits the scope of disclosure: “this information shall not include disclosure of the algorithmic code of the digital platform.” The law therefore emphasizes information, explanation, and human contact rather than code disclosure. It recognizes a “right to explanation” for decisions taken or supported by automated decision-making systems that significantly affect working conditions, requires access to a designated human contact person, and provides for a written statement of reasons within forty-eight hours for certain adverse decisions.
This differs from the Chilean approach. Chile’s Law 21.431 gives competent authorities, when required, a direct regulatory-access tool: access to the “programming of the algorithm” and “complete and sufficient explanations” of how it makes decisions. Uruguay, by contrast, expressly excludes disclosure of the algorithmic code while creating a worker-facing framework of information, explanation, human contact, and written reasons. These legal differences also operate within different enforcement settings: Uruguay’s workplace enforcement is centralized under the Inspección General del Trabajo y de la Seguridad Social within the Ministry of Labour and Social Security, while Chile relies on the Dirección del Trabajo, a technical agency linked to the Ministry of Labour and Social Security. The relevant policy question is therefore not which model is stronger in the abstract, but how each legal and institutional setting can make algorithmic transparency, explanation, and inspection effective in practice.
The Chilean implementation reports show why legal recognition is only the first step. The 2024 report states that “it was not possible to identify workers classified as dependent” and therefore “it is not possible to evaluate the effects of this distinction.” It recommends moving forward “as a priority” with the dissemination of information and inspection of “minimum and common elements” for dependent and independent workers. The 2025 report identifies more specific inspection limits: “It was not possible to conduct on-site inspections at the headquarters of platform companies” and, as a result, “it was also not possible to verify the functioning of the algorithms” in those premises. It adds that “infringements were concentrated in traditional labour matters” and that “only in 10 cases were sanctions applied based on the sector-specific regulation.”
The 2025 report also identifies practical bottlenecks: “undocumented foreign workers,” “subleased accounts,” and “the territorial dispersion of workers in relation to the legal domicile of the platforms.” It recommends “electronic inspection” to “access algorithms” and verify compliance with Law 21.431, as well as updated legal interpretations on “account renting,” “task-assignment mechanisms,” “performance evaluation through algorithms,” and “traceability of payment of remuneration/fees.” These findings point to a narrower implementation problem: the difficulty of verifying how platform systems operate in practice, even when the law recognizes algorithmic transparency and grants authorities access to relevant information.
Pulignano’s concept of algorithm governance helps frame why these legal and institutional differences matter beyond the formal text of each law. Her argument is that labour platforms adopt “a distinctive contractual form of employment” aligned with each country’s laws and regulations, while organizing service provision through algorithmic technologies. In that sense, platforms do not operate with a fixed model of algorithmic management; they may adapt their algorithmic governance locally in response to legal classification rules, enforcement institutions, labour-market conditions, and worker organization. For the Latin American cases, this suggests that the enforcement problem is not only whether the law recognizes transparency, explanation, human contact, or authority access, but whether labour institutions can observe how platforms operationalize those rules in practice through classification, task allocation, pay, evaluation, sanctions, and account status.
This is also where the ILO Convention creates an important opening. By treating automated systems and their impact on working conditions as matters of labour regulation, the Convention connects AI governance with labour rights rather than leaving algorithmic management only to data-protection or technology policy. The opportunity is not simply to add procedural transparency to platform work, but to strengthen the countervailing power that workers, unions, regulators, and advocates have been building through litigation, organizing, research, and policy debate over the last decade. Whether that opportunity becomes effective will depend on the same conditions highlighted by Chile and Uruguay: meaningful information, accessible explanation, human review, regulatory verification, collective representation, and inspection capacities adapted to digital work.