The Ghana Revenue Authority (GRA )’s latest attempt to boost revenue collection and close loopholes at the ports, through a new AI solution known as Publican, is already drawing comparisons to the controversial Strategic Mobilization Limited (SML) saga that once dominated public debate.
At the time, SML was introduced as a solution to “close gaps” in petroleum volume tracking. The “solution” was to reconcile data across government systems and ensure accuracy.
Yet, despite existing data integrations, the state paid millions to a third party to confirm what its own systems should have already verified. The outcome, critics say, was disappointing, leaving the government scrambling to recover losses.
The SML contract, after an expose’ by The Fourth Estate, is a subject of criminal prosecution by the Office of the Special Prosecutor (OSP) since the nation lost millions through the contract.
After the SML contract was abrogated, it appears a similar pattern of issues around the controversial deal seems to be emerging around the Publican AI solution deployed at the port.
Data and policy analyst, Alfred Appiah has drawn an interesting comparison between the two controversial initiatives, calling for more scrutiny and due diligence.
A Familiar Narrative
With Publican AI, the narrative is again about “closing revenue gaps”, but this time in trade. Authorities argue that while billions of dollars leave the country, declared values of imported goods are often lower, hinting at under-declaration.
The AI system is therefore positioned as a tool to detect and correct these discrepancies. However, the analyst raises a critical point that not all money leaving Ghana is tied to goods.
Payments for services, debt obligations, and other legitimate financial transactions also account for foreign exchange outflows. If these are not properly separated from goods-related transactions, the so-called “gap” may be overstated from the start.
“SML was reportedly “closing gaps” in petroleum volumes. Data from one government system, integrated via an API into another government system, somehow still required a third party to provide assurance that the two data sources matched. The state paid millions for this, and it all turned out to be a hoax. The state is now playing catch-up to recover the losses,” he indicated.
Early Results, Early Doubts
The GRA has already reported a near 50% increase in customs revenue within just two weeks of deploying Publican. On the surface, that sounds impressive.
But it also sounds familiar. SML once came with its own success claims, numbers that captured public attention before deeper scrutiny set in.
In the case of Publican, Appiah argues that a two-week window is too short to draw meaningful conclusions, especially in a sector where revenue flows can fluctuate due to shipment volumes and seasonal trade patterns.
Adding to the concern are complaints from industry players about possible overcharging, raising the possibility that higher revenues may not necessarily reflect improved efficiency, but rather inflated assessments.
“It’s unclear whether this revenue increase can be considered a reliable measure of success without a proper audit of the AI system’s assessments. But yet the public square is filled with this incredible success. SML had that too,” he remarked.
The Core Question: Why Another Layer?
Just as with SML, questions are being asked about the necessity of introducing a new system when existing infrastructure already exists.
Ghana’s customs operations are run through ICUMS, a platform designed to handle valuation, risk management, and compliance. If gaps exist, critics argue, should they not be fixed within that system rather than layering on a new solution?
It is a question that echoes the SML experience. Why pay a third party to validate or enhance what systems are already supposed to do?
Alfred Appiah maintained that, “I am yet to see any publicly available technical report demonstrating the need for this system to complement ICUMS. Is ICUMS not designed to perform these functions? And is Ghana Link not capable of incorporating “AI features” into ICUMS?”
In addition, he further flags the involvement of GRA in this current controversy. This is because GRA was very prominent in the SML, which ended up in a sage. For this reason, some former officials of the authority are under investigation for their alleged roles in the contract.
A Pattern or Progress?
The goal of improving revenue mobilisation is not in dispute. Ghana needs stronger systems to plug leakages and improve compliance. The concern is about how those solutions are designed, justified, and measured.
The SML episode showed how quickly promising interventions can unravel when transparency, clarity, and accountability are lacking. Publican AI, still in its early days, now faces a similar test.
Is this a genuinely transformative tool, or another case of overstated impact driven by early, unverified data?
