Does Your AI Customer Service Need to Disclose That It's AI? What Kenya's AI Bill Says
The Bill introduces a tiered, risk-based system for classifying AI systems, similar in spirit to approaches seen in the EU and parts of the US. Under this system, AI tools are sorted by how much potential harm they could cause, from minimal risk through to high risk.
Does Your AI Customer Service Need to Disclose That It's AI? What Kenya's AI Bill Says
If your business uses an AI chatbot to handle customer questions on WhatsApp, your website, or social media, a new question is about to matter a lot more than it used to: does that AI chatbot need to disclose its true nature to customers? Under Kenya's Artificial Intelligence Bill, 2026, transparency is increasingly mandatory, and it's a requirement worth understanding now before it becomes a compliance problem later.
The Artificial Intelligence Bill, 2026, sponsored by Senator Karen Nyamu, is Kenya's first comprehensive legislative attempt to regulate how AI systems are built, deployed, and used in the country, as TechCabal reported when the Bill advanced in early 2026 .
The bill classifies AI systems by risk level, and under this framework, something as ordinary as a basic customer service chatbot falls into the "limited risk" category, which carries a specific transparency obligation; disclosing to the user that they are interacting with AI.
This article breaks down exactly what the Bill says, what it means for a business running AI customer service today, and how a platform like Skyfalke Connect gives you the actual controls to build that disclosure into your chatbot from the start.
What Kenya's AI Bill 2026 Actually Says About Chatbots
Kenya has been building toward formal AI regulation for a while. The country's AI governance journey traces back to its National AI Strategy 2025-2030, tracked by the OECD's AI Policy Observatory as part of a five-year roadmap for AI adoption and governance, and that groundwork led directly to the Artificial Intelligence Bill, 2026, tabled in the Senate as Kenya's first dedicated AI law.
The Bill introduces a tiered, risk-based system for classifying AI systems, similar in spirit to approaches seen in the EU and parts of the US. Under this system, AI tools are sorted by how much potential harm they could cause, from minimal risk through to high risk. Chatbots fall into the lower end of that scale, but they are explicitly named, and they are not exempt from disclosure requirements.
Specifically, systems like basic chatbots are classified as limited or minimal risk, which still carries a lighter but real transparency requirement; telling the user they're interacting with AI rather than a person.
A related draft AI and Emerging Technologies Policy goes further, proposing that organizations be required to inform individuals whenever they're interacting directly with an AI system, particularly where that system could affect their access to services or their rights.
Who Does This Apply To?
The Bill distinguishes between a provider (the company that builds or sells the AI system) and adopters (the business that actually puts it to use with customers). If your business uses an AI chatbot, whether built in-house, bought off the shelf, or run through a platform like Skyfalke Connect, you are most likely an adopter under this framework, and the disclosure obligation sits with you as the business putting the tool in front of your customers.
Why This Is Happening Now
This isn't an isolated rule. It sits inside a broader shift in how Kenya is approaching AI governance, moving from voluntary strategy documents toward enforceable law. The AI Bill ties directly into Kenya's existing Data Protection Act, 2019, requiring that any AI system handling personal data comply with that Act as well, including rights like the ability to request human review of an automated decision.
A few things worth understanding about where things stand:
The Bill had not yet been signed into law as of its 2026 Senate deliberations, so businesses should treat this as a strong signal of coming requirements, not yet a fully enforced law.
Legal commentators have flagged real concerns about the Bill's broad scope and the compliance burden it could place on businesses, so the final version may shift from what's currently proposed.
Even so, the direction is clear enough that waiting until the law passes to think about chatbot transparency is a risky approach for any business already running AI customer service at scale.
What "Disclosure" Actually Looks Like in Practice
For a business running a chatbot, disclosure doesn't need to mean a complicated legal notice. In most frameworks like this, it means something straightforward: a customer should know, early in the conversation, that they're talking to an AI system rather than a human agent.
In practice, this usually looks like:
A clear opening message when a chat starts, identifying the assistant as AI-powered.
An easy, visible way for the customer to request a human agent if they'd prefer one.
Clear labelling if the AI is generating content, images, audio, or video, that could otherwise be mistaken for something made by a person.
A straightforward handoff process when the AI reaches the limit of what it can help with.
This is where the actual product decision matters. A chatbot that was built without any thought to disclosure can be reworked, but it's considerably easier when the platform you're using was designed to let you control the conversation from the first message onward.
How Skyfalke Connect Helps You Build This In
Skyfalke Connect's chatbot builder gives businesses direct control over how their AI assistant introduces itself, including the opening message a customer sees the moment a conversation starts.
That's the practical mechanism a business actually uses to meet a disclosure expectation like the one in Kenya's AI Bill: you decide what the bot says first, and you can make that first message explicitly state that the customer is chatting with an AI assistant.
Just as importantly, Skyfalke Connect's AI agents are built with human handoff in mind, so when a conversation needs a person, whether the customer asks for one or the AI reaches a limit, it can be routed to a live agent without the customer having to repeat themselves, since the full conversation history travels with them.
This matters across every channel a business uses. Whether your AI assistant is answering questions on WhatsApp, your website's live chat, or Instagram, the same disclosure and handoff principles apply, and Skyfalke Connect's unified AI platform is built to let you configure that consistently rather than channel by channel.
To be clear about what this does and doesn't mean: No software can guarantee compliance with a law that hasn't been finalized yet. What Skyfalke Connect does, it gives you the actual configuration tools, a controllable opening message, visible human handoff, and consistent behaviour across channels, to build a chatbot that's ready for this kind of requirement rather than one that would need a rebuild later.
For businesses thinking more broadly about AI governance and strategy beyond just chatbot configuration, Skyfalke's AI strategy team works with companies on exactly these kinds of readiness questions.
What Happens If You Ignore This
Kenya's AI Bill includes enforcement mechanisms, and non-compliance with disclosure and transparency requirements is explicitly treated as an offence under the proposed framework, not just a guideline businesses can quietly ignore. The Bill also proposes a public register of higher-risk AI systems operating in Kenya, part of a broader move toward bringing AI use out of the shadows and into regulatory view.
Even setting aside legal risk, there's a simpler business reason to get ahead of this: customers increasingly notice, and increasingly care, when they realise after the fact that they were talking to AI without being told. Being upfront about it from the start tends to build more trust than it costs.
Frequently Asked Questions
Is Kenya's AI Bill 2026 already law?
Not yet. As of its Senate deliberations in 2026, the Artificial Intelligence Bill remained a proposed law, not finalized legislation, though it represents a strong signal of where AI regulation in Kenya is heading. Businesses should prepare now rather than wait for final passage.
Does my chatbot really count as a "limited risk" AI system?
Under the Bill's proposed classification, basic customer service chatbots fall into the limited or minimal risk category, which carries a lighter transparency requirement than high-risk systems, but it still requires disclosing to users that they're interacting with AI.
What's the simplest way to disclose that a chatbot is AI?
A clear statement in the chatbot's opening message, identifying it as an AI assistant, along with an easy way for the customer to reach a human agent, covers the core of what's expected under proposed frameworks like this.
Does Skyfalke Connect make my chatbot legally compliant?
No single platform can guarantee legal compliance with a law that hasn't been finalized. Skyfalke Connect gives you the configuration tools, a controllable opening message, visible human handoff, and consistent behaviour across channels, to build disclosure into your chatbot, but the responsibility for compliance sits with your business as the adopter of the AI system.
What other Kenyan laws apply to AI customer service beyond the AI Bill?
The AI Bill is built on top of Kenya's existing Data Protection Act, 2019, which already governs how personal data is collected, processed, and stored. Any AI chatbot handling customer data needs to be considered against both frameworks, not just the AI Bill alone.
Conclusion
Kenya's AI Bill 2026 signals a clear direction: AI customer service tools, even simple chatbots, are expected to tell customers they're talking to AI. The law isn't finalized yet, but the businesses that prepare now, by choosing a platform that gives them real control over how their AI assistant introduces itself and hands off to a human, will be in a far stronger position than those scrambling to retrofit disclosure later.
If your business is running or planning an AI chatbot, book a Skyfalke Connect demo to see how the chatbot builder gives you control over exactly how that first conversation starts.