Kenya's State Department for ICT and the Digital Economy published a draft national artificial intelligence policy on 24 July, open to consultation until 4 August, committing the government to publish a statutory Fair Pay Reference Framework for data-annotation and content-moderation workers, alongside duty-of-care standards covering exposure to harmful content, mental health support and contracting terms.1
The draft describes Kenya's place in AI supply chains as concentrated in lower-value segments such as data annotation and content moderation, with limited structured pathways for progression into higher-value roles. Governments rarely write down that their comparative advantage is being cheap, and the sentence is doing policy work: a pay framework needs a stated problem before a ministry can defend the price it sets.
Kenyan annotators earn between $1.46 and $3.74 an hour on figures relayed by the trade outlet WeeTracker, against $21 to $27 for comparable American work.2 Annotation is the labour that produces the training data and the safety filtering behind consumer AI products, which makes it the one segment of this industry where wage regulation would touch the cost base of the model developers directly.
A statutory floor sets a price that a buyer must meet or leave; a reference benchmark publishes a figure and hopes procurement Teams read it. Nothing in the draft yet settles which of the two Kenya intends. India's outsourcers spent the same quarter arguing over graduate intake instead, with Wipro adding 888 net staff against its own zero-fresher pledge, and no high-income government proposed a price floor for the human labour that trains the models at all.
