Over 30% of health insurance claims in Kenya could be fraud-related. Not suspected; could be. And according to the leaders managing those claims every day, the number is still growing.

‍

That figure came from Njeri Jomo, CEO of Jubilee Health Insurance Kenya, during Curacel Frontline #1, a panel discussion that brought together senior insurance executives from across East Africa to talk about something the industry has been slow to say out loud: fraud in health insurance is no longer an occasional problem. It has become structural. And the insurers winning the fight are the ones who stopped treating it like one.

‍

The problem got smarter before the solutions did

For years, health insurance fraud looked predictable: impersonation, duplicate submissions, the occasional bad actor slipping through. Rule-based detection systems worked well enough because the fraud itself was episodic. Detectable. Containable. That has changed fundamentally.

‍

"We've come from a situation where it would be impersonation here and there to something a lot more systemic," said Jomo. "Collusion is often one of the harder things to pick up."

‍

The shift is driven by incentives. Post-COVID, providers moved aggressively to grow revenue through higher incidence rates and increased average costs, a direction that works directly against insurers trying to manage the same metrics. More critically, the rule-based systems designed to catch fraud became predictable. Providers learned how they worked and adapted. "Providers know how you prefer your documents received," Jomo noted. "They realise you like it done this way and that way, and the pass rate is higher." Static rules created a ceiling. Intelligent systems remove it.

What happens when AI enters the claims process

The gap between rule-based detection and AI-powered adjudication is not incremental; it is structural. Dennis Twambale, Head of Health Operations at Jubilee Insurance Uganda, laid out what that gap looks like in practice.

‍

"Tens of thousands of claims which would have been done in over 20 days are done in under a day, or a couple of hours," he said. "Over 80% of our claims are going through AI."

‍

The operational shift extends well beyond speed. Provider reconciliation, historically one of the most drawn-out processes in claims management, taking anywhere from three to six months, now completes in under a month. Incurred claims have dropped by approximately 8% as providers align more closely with expected billing standards, not because they were confronted, but because they were educated. The intelligence flowing back to providers through AI-powered claims systems is changing behaviour upstream. "That feedback is helping improve healthcare outcomes," Twambale said. "You're streamlining the quality of care."

‍

This is precisely the kind of transformation Curacel Health AI is built to enable; giving health insurers across Africa the infrastructure to move from manual, reactive claims management to real-time, AI-driven adjudication that catches fraud, reduces leakage, and improves provider relationships simultaneously.

The barriers that slow everyone down

Despite results like these, adoption remains uneven. Dr. Benjamin Otieno, GM of Risk and Compliance at GA Insurance Kenya, pointed to trust as the most fundamental obstacle. "AI is a black box," he said. "We are passing into it a lot of data and it is churning out decisions. Whether those decisions are biased or not, that is a very valid concern."

‍

The auditability problem compounds this. Conventional auditors are not yet equipped to interrogate AI systems, and clinicians resist having their judgement questioned by an algorithm. But Otieno was clear that this is a solvable problem, it requires transparency in how models are trained, ongoing bias correction, and systems that can explain their decisions. The tools exist. The willingness to demand them from technology partners is what varies.

‍

Abdul-Jabbar Momoh, VP Commercial at Curacel, raised a dimension that often gets missed entirely. "Insurers come to us wanting to layer intelligence for faster adjudication," he said. "But when you dig deeper in discovery, you begin to see broken points from poor data quality, to fragmented intake workflows and inconsistent provider relationships. Sometimes you have to fix the foundation before AI can deliver real value." The lesson is not that AI is too complex to implement. It is that the quality of what goes in determines the quality of what comes out. Curacel Health AI works with insurers at exactly this stage, not just deploying a product, but building the operational foundation that makes it work.

Beyond fraud: what the best insurers are already doing

The most forward-looking part of the conversation was not about catching fraud at all. It was about what becomes possible once the foundation is in place. Jomo described Jubilee's shift toward value-based care using claims intelligence to challenge treatment protocols, question prescriptions, and drive outcome-focused conversations with providers. "When we remove the financial aspect, it's really about a human being. Are they genuinely getting better?"

‍

Momoh mapped out where the market's early adopters are already headed: predictive underwriting, preventive care programs, personalised health products designed around actual member behaviour. "For our earliest adopters, they're now moving to rely on that intelligence to be more predictive. How do you move from being preventive to being proactive?"

‍

For Twambale, the most durable change has been in how providers and insurers relate to each other. "Opposed to fighting providers, educate them. Connect the dots. Show them how it affects the entire ecosystem, then the relationship becomes much better."

‍

That shift from adversarial to collaborative, from reactive to predictive, from managing claims to managing health is where the conversation in East Africa is heading. The infrastructure is now in place. The question is what you build on top of it. If you are an insurer thinking about where to start, or an operator ready to go further, talk to the Curacel team.

‍