03/06/2025
Legal Implications of Agentic and Generative AI Systems
Artificial Intelligence is revolutionising industries across Kenya. From banks to logistics firms, AI is enhancing productivity and unlocking new value. Most Kenyans are already familiar with generative AI—like ChatGPT—which can write content, generate images, or produce code. But an equally significant innovation is agentic AI: systems that not only generate content but also act on it autonomously to pursue specific goals.
Generative AI is like a smart intern: you ask, and it creates. Agentic AI is more like a junior manager: you define the objective, and it plans, executes, and follows up—without constant prompting. For example, a generative AI may draft a contract, but an agentic AI might send it out, follow up, and confirm the terms. The difference is real-world action and autonomy.
This distinction has critical legal implications. In Kenya, generative AI raises questions of intellectual property, bias, and data protection—risks that are often mitigated by human review. Agentic AI, on the other hand, creates conduct-based risks. If it autonomously books appointments, executes payments, or processes personal data, the liability can shift from human error to algorithmic misconduct.
Firms operating in Kenya must take compliance seriously. The Data Protection Act, 2019 and related Regulations classify some agentic AI tools as data processors. This triggers obligations on privacy impact assessments, contractual safeguards, and data subject rights. Further, agentic AI systems must be treated like employees with delegated authority: they require clear scope, escalation paths, and internal audit trails.
As AI adoption grows, every Kenyan business must ask: does this system merely generate or does it also act? The answer defines your legal exposure. Sensible governance, ongoing training, and robust contractual terms are now non-negotiable. AI can drive national innovation—but only if we govern it responsibly.
