Artificial intelligence is transforming society, particularly work environments, with the rise of click work complicating labour rights. AI companies contract third-party firms in Kenya to train datasets, aiming to reduce bias and toxicity in AI models, under the guise of creating ethical AI. Despite AI systems being computer-based, they fundamentally rely on human labour for training datasets and for algorithm development. These workers could be content moderators or data labellers.
A TIME investigation revealed concerns of working conditions for AI labourers, including low wages, toxic environments causing mental health issues, insufficient compensation, and stressful working conditions. Kenya's status as a regional ICT hub has made it a centre for this work, including dirty work, with several workers recently sharing their challenges in interviews with 60 Minutes Australia.
Platform cooperatives' response: Opportunities and gaps
In Kenya, individuals and organizations have mobilized to improve AI governance and advance platform cooperativism in response to these challenges faced by AI labourers.
The cooperativism movement has gained momentum through strategic meetings and formal organization. An 'International Solidarity Meeting' in Nairobi, Kenya's capital, in August 2024, brought together content moderators and data labellers to articulate their demands against AI companies and contractors. Their demands included addressing discrimination, ensuring fair working conditions, accountability for worker safety, and recognition of data work as a legitimate occupation.
February 2025 marked a significant milestone with the launch of the Data Labellers Association in Kenya. The association attracted over 300 members on its first day. The association has established an interactive website to advocate for fairness in Kenya's AI labour market, describing itself as standing with 'often-overlooked workers behind AI.' The association created formal membership structures with an elected President serving as its spokesperson, demonstrating its commitment to challenging systemic injustices, including labour invisibility and lack of recognition in national frameworks.
Strategically, the association plans to negotiate collective bargaining agreements with AI companies. It has already formed connections with the African Content Moderators Union to leverage multi-level cooperation.
The cooperativism efforts have been emboldened by recent media spotlights. Media organizations have played a crucial role in highlighting AI labourers' challenges, with CBS and 60 Minutes Australia producing investigative stories about their experiences.
Civil society organizations have also facilitated multi-stakeholder dialogues, workshops, and roundtables to voice labourers' concerns and engage contracting companies about how AI labour practices affect workers' wellbeing. Some of the forums often include legislators and government officials, promoting practical policy interventions.
Research sponsorship has further strengthened cooperativism efforts. The Pulitzer Centre for AI Accountability regularly funds journalists to conduct accountability research in the AI sector. In November 2024, the Centre collaborated with AI Kenya to organize a meeting that highlighted labourers' challenges and explored pathways to fairness. These multidisciplinary gatherings emphasize holistic approaches to healing and finding restorative justice, using existing frameworks to help affected workers voice their concerns.
Through formal associations, media advocacy, multi-stakeholder engagement, and legal action, Kenya's AI labor cooperativism movement has created structures to address labour invisibility, promote fair working conditions, and pursue accountability from major technology companies employing AI labourers in the country. These highlighted mainstream and complementary cooperativism movements have realized some success. For example, the African Union of Content Moderators has successfully filed cases challenging Meta's impact on workers' mental health. A recent court ruling establishing Kenyan courts' jurisdiction over Meta's operations in Kenya represents a promising trajectory that will augment the strategies deployed by these cooperativism movements.
Politics of the efforts and creation of safe havens
The decent cooperativism efforts, however, face constant challenges. The challenges arise from the threat by the contracting companies to relocate to neighbouring countries with less worker agitation and minimal government intervention. For example, Sama has threatened to relocate to Uganda, a country neighbouring Kenya to the East. Even though the threats of moving to such jurisdictions may never be realized, the threat of moving alone may be enough regulatory concern for two main reasons.
One, the threat of business flight goes hand in hand with potential negative implications on job loss and economic development. Thus, it is inevitably laced with an undertone for invoking political interests. Such threats of relocation carry with them a consequence of giving Uganda an upper hand over Kenya, at least better than the status quo, as the hub for technology investment in East Africa. The companies usually leverage the possibility of such a move as a tool for negotiating with political actors to preserve the status quo of worker exploitations that currently form the basis for agitations. Where they are unsuccessful in their negotiation or where the hosting State takes stern regulatory measures, they can then make good their threats by shifting jurisdictional locations.
This first undertone defeats the cooperativism effort by promoting lobbying. For example, as a reaction to the platform cooperativism efforts, the AI companies whose significant investments cannot be downplayed have invoked dynamic of work and economic development as a basis to lobby some sections of the political class to insulate themselves from claims arising from the AI worker challenges. So much so that some top political brass in Kenya seemed to have been influenced by such lobbying. For example, the Kenyan President even publicly hinted at the possibility of changing the law to prevent suits from being instituted against AI contracting companies in Kenya.
The proposed drastic step of taking away court jurisdiction over AI companies, assuming it were possible as the Kenyan President suggested, would mean that AI companies and their contractors would be at liberty to operate in Kenya while evading potential legal liabilities for the woes and challenges of AI labourers. Such a possibility would cause Kenyan jurisdiction to be safe havens for the AI companies, enabled by contemporary political machinations.
Two, such a shift is also laced with an undertone of safe havens caused by comparatively weaker legal frameworks and a lesser environment for activism in Uganda. By singling out Uganda as a jurisdiction of preference, should the cooperativism efforts in Kenya persist, AI companies were giving a signal that Uganda is or would be a safe haven where they will not be 'troubled' as much as in Kenya.
Challenges with the safe havens
The safe havens present real challenges whose impacts transcend beyond the impacts of the AI worker.
Firstly, the AI developing or manufacturing companies or their contracting partners may take advantage of these safe havens to strategically choose, as the State of principal operation, the country with the least compliance standards. This may affect the realization of standards for ethics as well as rights and other fundamental freedoms.
Secondly the safe havens create an environment where AI companies and their contractors can imagine all possible ways of organizational restructuring and supply chain planning to avoid liability for harm caused to AI labourers.
At the organizational levels, safe havens in AI work may arise from contractual arrangements involved in the AI work. For example, it would be possible that the contracting company, acting on behalf of the AI company, could be in another jurisdiction besides the national jurisdiction where the workers are physically located. That means that digital cooperativism efforts in the jurisdiction where the workers are located, though plausible, may not be enough.
These challenges may have cross-border implications. They are, therefore, an affront to the State-specific platform cooperativism efforts and State-centric programmatic interventions in Kenya.
Conclusions on lessons learned
Overall, it is clear how the politics of AI work could lead to the creation of safe havens. The creation of AI safe havens is antithetical to the idea of digital cooperativism. The use by AI companies of threats of movement to countries that could serve as safe havens is a tool for negotiating the maintenance of the status quo, which the emerging digital cooperativism structures are persistently challenging.
Furthermore, the Kenyan AI worker situation offers insights into platform cooperatives and the concept of 'platform co-op safe haven' within sub-regional blocs. As East African Community members, Kenyans' experience reveals that companies can leverage relocation threats due to the lack of a unified sub-regional regulatory approach.
Comments
Post a Comment