New ILO convention recognizes AI as a tool for labor management and sets a historic precedent

Between 1 and 12 June, Geneva brought together representatives of governments, employers and workers from around the world at the International Labour Conference (ILC) to resume discussions on one of the key issues from the previous session: the draft convention and recommendation on decent work in the platform economy. The adoption of the instrument would make it the first international legally binding instrument aimed at recognizing and regulating the exercise of the right to work in the context of new technologies. After grueling days of negotiations, a historic milestone for the sector was reached: the convention was adopted!

In this column, we will take stock of the discussions surrounding the adoption of Convention No. 193, highlighting the main demands put forward by workers’ organizations, as well as the nuances and tensions surrounding them in relation to the interests and positions of governments and the private sector. We will focus on the provisions on the algorithmic management of digital platform work, as their inclusion in this instrument represents one of the major achievements of the negotiation process. Indeed, since the expansion of this form of work, the impacts of algorithmic systems on platform workers’ living and working conditions have been one of the central issues driving their struggle.

A convention to enshrine and advance rights

In previous publications, we have explained the significance of the International Labour Conference (ILC) and the conventions and recommendations discussed and adopted there for the world of work. It is worth recalling that International Labour Organization (ILO) conventions are intended to establish minimum standards that guarantee decent working conditions and fundamental rights for all workers, across all sectors, regardless of their contractual status. Once an ILO member State ratifies one of these legal instruments, it is required to incorporate its provisions into national legislation and practices.

Among the demands put forward by workers and incorporated into the discussions at the International Labour Conference (ILC) in 2025 were meaningful accountability in algorithmic decision-making; transparent wage and task-assignment structures; guarantees that work-related costs are covered by the companies; and protection against the misclassification of workers as independent contractors or self-employed. They also called for the establishment of the principle of the primacy of facts; transparency in performance evaluation processes, as well as in the metrics used and their impact on future employment opportunities; protection against health and safety risks arising from platform work; safeguards for privacy and clarity regarding the use of personal data; and guarantees for collective bargaining and freedom of association.

The groundwork for these demands can be found in the lists of demands submitted by labor unions around the world, which helped shape an international standard even before the 2025 ILC. The agenda advanced by platform workers’ organizations and their remarkable capacity for collective action emerged as a key driving force behind the adoption of the new standard.

Adopted with 406 votes in favor and 8 against, Convention No. 193 applies both to workers on web-based platforms and to those on location-based platforms, such as transport and delivery services, thereby expanding its scope of protection compared to earlier versions of the text. The instrument establishes that workers must be correctly classified, based primarily on the facts surrounding the performance of the work rather than on corporate or subjective interpretations of their independent status. It also covers and recognizes the promotion and respect of freedom of association and trade union freedom, as well as the right to collective bargaining; includes the right to refuse dangerous work without retaliation; and guarantees the right to protection against violence and harassment, including when perpetrated online or involving clients.

The new instrument also reaffirms that digital platform workers have the right to request access to, correction of, and deletion of the personal data processed by these applications; prohibits the suspension or deactivation of accounts or the termination of employment or contracts based on discriminatory grounds; calls for human intervention and review in algorithmic decision-making aimed at withholding payments, suspending or deactivating work accounts, or ending employment relationships; and requires technology companies to provide information, prior to hiring, regarding the use of automated systems and their impact on working conditions.

The formal recognition of these rights, as well as the establishment of principles for determining and understanding the nature of the employment relationship in these forms of work, are critical in a context where the paradox of supposed flexibility and autonomy in these jobs persists, while unjustified and widespread suspensions and deactivations of work accounts on location-based platforms continue to be documented.

Why, then, is the inclusion of provisions on the algorithmic management of work so significant?

Algorithmic management as a source of tension

During the discussions, the employers’ group representing the companies took a particular stance on the articles addressing the use of automated systems. According to them, “regulating technology” would not fall within the ILO’s mandate, and doing so could hinder innovation and economic growth, while also constituting an intrusion into the operations of private companies and into commercial information protected by intellectual property rights. Their arguments were supported by some governments, including the United States.

The proposal put forward by this group to exclude algorithmic management from the ILO’s scope of action would mean leaving the rules governing how algorithms assign tasks, set compensation, or deactivate accounts in the hands of corporate policies and opaque technical designs that do not necessarily take into account workers’ recognized rights. In this regard, the principles of human oversight and access to information adopted in the convention are essential to ensuring effective safeguards against the use of algorithms in this type of work.

However, while the final version of the instrument manages to address some of the workers’ concerns and demands, its final section broadly states that “In implementing this Convention, each Member shall take appropriate measures to protect commercially sensitive information of digital labor platforms.” If applied to the use of algorithms, something already invoked by some governments worldwide and defended by companies at the regional level, this provision reopens tensions addressed in multiple analyses regarding the nature of algorithms when they make important decisions, such as those mentioned above. These tensions culminate in a key question: Is algorithmic management merely for market allocation, or does it involve the supervision of work?

A technology that undermines rights

According to workers’ organizations, algorithmic management cannot be regarded as a neutral market tool because, in practice, it influences wages, working time, and disciplinary measures—areas traditionally governed by labor law. Allowing States to limit the guarantees established in the convention in the name of protecting commercially sensitive information has significant implications for workers’ rights and working conditions. Although the final provision appears to suggest a conflict between fundamental rights that would require States to assess the necessity, proportionality, and reasonableness of protecting intellectual property in relation to labor rights, this was not explicitly stated in the text and may leave room for broad interpretations of the protections afforded to corporations.

Consider this for a moment: when a person supervises, assigns tasks, and evaluates performance, this is recognized as management and is subject to labor protections. Workers, therefore, can demand fair remuneration, safe working conditions, and due process guarantees. Why, then, do companies claim that when an algorithm exercises the same level of control, it is not management at all, but simply a feature of digital service design? The consequences are clear: algorithmic systems have been shown to assign microtasks in a matter of seconds, continuously rank workers, impose sanctions, and even block access to the platform, often with little or no explanation.

Arguments regarding the trade secret status of algorithms in these forms of work have already been the subject of judicial decisions in our region. By introducing a safeguard in its implementation provisions to protect the commercially sensitive information of digital labor platforms, the Convention may allow this principle to be invoked to shield the opaque algorithms that underpin labor management. This lack of transparency is not incidental; it is inherent to the business model. Leaving undefined what should be considered commercially protectable, without reinforcing the requirement to assess its necessity, proportionality, and reasonableness, could mean that millions of workers will have to struggle even harder to protect themselves against the deployment of AI in the platform economy. The fact is that the ILO’s conventions carry considerable global weight, as governments often rely on them as a guide for labor standards and policies.

An Ongoing Struggle

Convention No. 193, as a landmark instrument for platform work, sought to strike a delicate balance: establishing principles broad enough to adapt to the sector’s changing realities while still offering concrete safeguards; and expanding labor rights without compromising the operational flexibility of business models. Likewise, the accompanying Recommendation, to which various provisions from the drafts were moved as a result of the tripartite consultation process, ultimately remained pending.

The newly adopted instrument will not operate in isolation in the countries that decide to ratify it. Efforts to regulate these forms of work are already underway around the world, particularly in Latin America. In our region, five countries have regulations governing location-based work on digital platforms. In this regard, the Convention will interact with States’ existing regulatory frameworks, both in relation to the platform economy and to laws concerning personal data protection, freedom of association and collective bargaining, artificial intelligence, access to information, competition policy, and intellectual property.

The consolidation and expansion of the progress achieved through this instrument will require a coordinated and sustained effort by trade unions, civil society organizations, States, and the private sector. Workers’ organizations will now focus their efforts on encouraging governments in their countries of origin to ratify the Convention. This is essential because, although the instrument establishes provisions on remuneration, automated systems, personal data, and health and safety, it does so through a principles-based approach that largely leaves the ultimate responsibility for its implementation to national legislation and practices. The way representatives from Latin American governments, including Argentina, Chile, Paraguay, Panama, and Costa Rica, voted —by abstaining from voting on the Convention— already points to a scenario in which trade unions will need to undertake special mobilization efforts to achieve the expected ratification.

In addition, it will be essential for civil society to work alongside governments and recommend, with regard to the algorithmic management of platform work, tools that ensure the effectiveness of the measures set forth in the Convention. One such tool could be the creation of public registries of the automated systems deployed on each platform, containing information on their purpose, design, and operation. Another possible tool is the development of human rights impact assessments focused on workers, examining the impact of these systems on wages, working hours, health and safety, and freedom of association. These assessments should be developed in collaboration with trade unions and organizations representing workers.

Algorithms should never have the final say over working conditions, the right to freedom of association, or a worker’s future. The right to challenge automated decisions must be guaranteed, as must the right to request human review and seek remedies through labor law mechanisms.