GOVERNMENT MOVES TO TIGHTEN CONTROL OVER AI WITH NEW 2026 BILL

Kenya is stepping into a new and controversial phase of digital regulation following the introduction of the Artificial Intelligence Bill, 2026, sponsored by Karen Nyamu. The proposed law signals a clear shift: the government is no longer just observing artificial intelligence—it wants structured control over it.

At the center of the bill is the creation of a powerful new office—the Artificial Intelligence Commissioner. This office would be nominated by the President and approved by Parliament, giving the Executive significant influence over who ultimately oversees AI regulation in the country.

And this is where things get serious.

The Commissioner wouldn’t just be a symbolic figure. The bill hands them sweeping powers: entering premises to inspect AI systems and data, compelling companies to hand over documents, issuing enforcement directives, and imposing fines. They would also have authority to summon individuals to testify and even classify AI systems based on the level of risk they pose.

That classification power is not small. If a system is deemed “unacceptable risk,” it is outright banned in Kenya. No negotiation.

On paper, this sounds like responsible governance—especially in an era where AI can influence elections, financial systems, and even public opinion. But look closer, and the concern becomes obvious: who defines “risk,” and how objective will that process be?

The bill also gives the Commissioner authority to set ethical standards, security protocols, and bias detection guidelines. Again, reasonable in theory. But in practice, this creates a centralized gatekeeper for innovation.

If you’re building AI in Kenya, you’re no longer just solving a technical problem—you’re navigating a regulatory minefield.

The proposed office would run conformity audits, conduct risk assessments, and monitor AI systems even after they’ve entered the market. It would also maintain a public register of high-risk AI systems, including those used by county governments, and investigate complaints ranging from bias to rights violations.

There’s also a softer side to the bill—at least on the surface. It introduces regulatory sandboxes meant to allow innovators to test AI systems in controlled environments. That’s a smart move if implemented properly. But don’t ignore the contradiction: the same authority enabling innovation is also the one with power to shut it down.

An advisory committee is also proposed, bringing together representatives from government agencies, private sector players, and civil society. Sounds balanced—but let’s be honest, advisory committees rarely override real power. The Commissioner still holds the decisive authority.

For companies dealing with high-risk AI, the burden increases significantly. They’ll be required to submit annual compliance reports, with parts of that information made public. That introduces transparency—but also exposes businesses to scrutiny and potential misuse of disclosed data.

And then there are the penalties.

Non-compliance could cost up to KSh 5 million, two years in prison, or both. That’s not a slap on the wrist—that’s meant to enforce obedience.

So what are we really seeing here?

Strip away the legal language, and this bill is about control versus innovation. The government is trying to get ahead of AI risks—but it’s doing so by centralizing power in a way that could easily be abused or misapplied.

Here’s the blunt truth:
If implemented poorly, this law won’t just regulate AI—it will slow it down, scare off investors, and push serious developers to operate elsewhere or underground.

If implemented well, it could position Kenya as a leader in ethical AI governance in Africa.

But that outcome depends entirely on execution—and Kenya’s track record with regulatory bodies isn’t exactly spotless.

Right now, this bill is less about what AI is, and more about who gets to control its future in Kenya.