Kenya has unveiled a draft policy to introduce labour protections for workers supporting artificial intelligence (AI) systems, placing greater responsibility on global technology companies and outsourcing firms operating in the country.
The proposed AI worker protection policy sets minimum standards for pay transparency, mental health support and workplace conditions for content moderators, data annotators and AI quality evaluators.
If adopted, the policy would significantly expand labour protections across the AI value chain in Kenya.
The draft, published by the Ministry of Information, Communications and the digital Economy as part of a broader Artificial Intelligence and Other Emerging Technologies Policy, is open for public consultation until 4 August.
John Tanui, principal secretary for the State Department for ICT and the Digital Economy, said the policy reflects the balance Kenya is trying to strike between embracing AI and governing it responsibly.
“Artificial Intelligence will shape the next era of economic growth, public service delivery and global competitiveness. The question is not whether Kenya should embrace AI, but how we govern it responsibly while unlocking its full potential,” Tanui said.
He said the draft policy is an opportunity for stakeholders, including innovators, researchers, academia, industry, start-ups, civil society and development partners, to shape a framework guiding AI development, adoption and governance.
“Together, these contributions will help shape a trusted, inclusive and innovation-driven AI ecosystem that advances Kenya’s digital transformation and global competitiveness,” he said.
The policy calls for occupational protection guidelines covering written employment contracts, psychosocial support, grievance procedures and minimum workplace standards.
It also proposes duty-of-care obligations requiring employers to protect workers from prolonged exposure to disturbing material and provide access to mental health services.
The framework applies to both local AI companies and international firms operating through outsourcing partners, extending accountability to global technology companies commissioning the work.
A central proposal is a fair-pay reference framework benchmarking wages for data annotation, content moderation and AI quality evaluation against international rates, with employers required to disclose pay structures through compliance reporting.
The move follows years of complaints from Kenyan workers employed by outsourcing firms serving companies such as OpenAI and Meta, with some content moderators reportedly earning between $1.46 and $3.74 an hour, compared with $21 to $27 an hour for similar roles in the US.
The draft also addresses workplace surveillance, calling for monitoring that is proportionate and respectful of workers’ rights, alongside stronger grievance mechanisms.
Content moderators review and remove material that violates platform rules, while data annotators label data to improve AI system performance.
Workers have said the job exposes them to graphic and harmful content, often without adequate psychological support. A KICTANet-affiliated study found that 27 per cent of surveyed moderators reported significant mental health distress, including anxiety and post-traumatic stress.
Kenyan courts have addressed related disputes. The Court of Appeal rejected Meta’s argument that Kenyan courts lacked jurisdiction, allowing content moderators to pursue claims locally.
A separate ruling also ordered Meta to provide appropriate medical, psychiatric and psychological support to moderators and annotators.
Share
