Kenya's ICT Ministry has unveiled a draft policy aimed at regulating artificial intelligence (AI) models operating within the country, even if the companies behind them lack a physical presence in Kenya. This move targets international AI providers such as OpenAI, creator of ChatGPT, and Meta, whose AI technologies are increasingly integrated into Kenyan businesses and public institutions.

Extraterritorial Jurisdiction for AI Oversight

The proposed framework adopts an extraterritorial jurisdiction model, similar to the European Union's approach to data protection and technology regulation. This allows Kenyan authorities to hold foreign AI firms accountable if their products or services impact Kenyan users or public interests.

Scope and Requirements

  • The policy covers a broad range of stakeholders including AI software vendors, cloud and compute service providers, data intermediaries, and public sector technology suppliers.
  • AI systems will be classified by risk level, with high-risk applications—such as those in healthcare, education, law enforcement, and elections—subject to stricter oversight.
  • Tech companies will be required to conduct risk assessments, ensure transparency with explicit labelling of AI-generated content, implement human oversight, and comply with cybersecurity standards.
  • Vendors must disclose details on data sources, model limitations, auditability, and mechanisms for redress.
  • International firms bidding for government AI contracts will need to partner with local technology companies.
  • Public institutions must perform AI impact assessments before deploying high-risk systems and maintain a public register of AI deployments, barring national security exceptions.

Labour Protections and Worker Welfare

The draft policy also introduces protections for Kenyan AI content moderators and data annotators working for outsourcing firms. It proposes minimum contract standards, access to mental health support, and fair pay aligned with international benchmarks, addressing previous concerns about poor working conditions and psychological strain.

Enforcement and Penalties

While specific penalties and regulatory obligations are yet to be detailed, the policy draws inspiration from the EU’s enforcement model, which includes substantial fines and requirements for foreign companies to appoint local representatives for regulatory compliance.

This initiative marks a significant step by Kenya to assert control over AI technologies influencing its digital ecosystem and protect the rights and interests of its citizens.