Stakeholders push for data sovereignty as AI conference tackles power shortages, worker exploitation

Regulators, policymakers, and industry stakeholders converged at the Hyatt Regency in Westlands for the three-day Africa AI Stakeholder Meeting, confronting the physical, economic, and human bottlenecks threatening the continent's ambition to scale artificial intelligence.

Held under the theme "AFRICA'S AI ADVANTAGE: INNOVATIVE GOVERNANCE FOR AN ERA OF SCALE," the summit took place as Kenya actively positions itself as East Africa’s primary digital hub, even as global tech giants and researchers continue to debate safety risks and call for development slowdowns.

Moving beyond high-level policy frameworks, the gathering tackled immediate operational challenges: strict data protection compliance, severe compute infrastructure deficits, worker exploitation in data labeling, and the protection of indigenous intellectual property.

Addressing compliance enforcement, John Walubengo, OGW, Deputy Data Commissioner for Compliance at Kenya’s Office of the Data Protection Commissioner (ODPC), outlined how strict regulatory oversight is yielding concrete administrative benefits.

Walubengo highlighted a recent Memorandum of Understanding (MOU) between the Central Bank of Kenya (CBK) and the ODPC, which requires digital lenders to obtain an ODPC compliance certificate prior to receiving CBK operational licensing.

The arrangement establishes a regulatory blueprint for how African nations can enforce baseline data protections across corporate sectors. Regional counterparts, including Dr. Emmanuel L. Mkilia of Tanzania and Ibunkunoluwa Owa of Nigeria, echoed the call for cross-border data interoperability and co-created regulatory frameworks designed to balance market growth with citizen protection.

The summit also surfaced sharp warnings regarding the hidden labor supporting global AI systems. Joan Kinyua, President of the Data Labelers Association, called on international technology firms to address severe mental health challenges faced by African workers tasked with training foundation models.

Kinyua cited high rates of psychological trauma among local data labelers exposed to explicit and disturbing material. She noted that strict non-disclosure agreements (NDAs) enforced by foreign AI developers exacerbate the issue by legally preventing workers from seeking professional emotional support or discussing their conditions with colleagues and family.

According to reporting by TIME Magazine, companies that have utilized human workers in Kenya for AI training and data annotation include Sama, OpenAI, Remotetasks, and several Chinese tech firms.

Concerns over labor practices extended into broader debates on intellectual property and data extraction. Panelists including Linda Musita of GIZ Kenya and Lydia Kila Taban of Masakhane argued that African creators and linguistic communities must retain control over how their cultural assets and local languages are ingested into generative AI models.

Speakers called for the immediate establishment of durable consent mechanisms and standardized metadata tracking to prevent foreign entities from unilaterally extracting African data assets without fair compensation or attribution.

Despite ambitions for technological self-reliance, the summit repeatedly turned to Africa's severe infrastructure limitations. Dumisani Dube of the AI Skills & Compute Africa Foundation (AISCA) questioned the sustainability of relying on foreign processing infrastructure, urging delegates to address who ultimately controls dataset storage and computing resources.

Faith Waithaka, cloud and service provider segment lead at Schneider Electric, noted that Kenya currently accounts for just 0.6 percent of global data center capacity, with existing local facilities operating at an average utilization rate of only 37 percent.

These constraints come amid news that Microsoft’s proposed $1 billion geothermal-powered data center in Olkaria, Kenya, was suspended over energy capacity and financial guarantee concerns. President William Ruto previously acknowledged that powering the 1-gigawatt facility at full capacity would require "switching off half the country," given Kenya's total grid capacity stood at 3,840.8 megawatts as of June 2025, according to the Energy and Petroleum Regulatory Authority (EPRA). Energy specialists Winfred Mutinda, Denis Nderitu, and Alfred Osiko emphasized that while data centers represent a key economic opportunity, infrastructure expansions must be strictly aligned with regional power availability, water resources, and actual demand.

Closing out the three-day agenda, delegates focused on translating diagnostic findings into actionable policy options, government commitments, and a multi-year work program targeted through 2027. Pauline Kanana, Managing Partner at IDI, urged regulators, universities, and private sector leaders to construct clear, adoptable implementation pathways.

Addressing questions regarding the concentration of AI power globally, Didier Nkurikiyimfura, Director of Emerging Technologies and AI at Smart Africa, stressed that AI should only serve the continent if African nations are full partners in its development rather than external consumers.

Nkurikiyimfura advocated for shared regional infrastructure, cross-border access agreements, and local financing structures to prevent African startups and talent from migrating to Silicon Valley.

Academic leaders, including Dr. Timothy Kamanu, Head of the Division of Data Science at the University of Nairobi and founder of DASCLAB, reaffirmed that multidisciplinary research and local computational problem-solving will be central to securing the continent's digital independence.