The ongoing controversy surrounding Ghana’s Publican AI system offers the country something more valuable than a cautionary tale; it presents a live, urgent opportunity to reassess how artificial intelligence is introduced into critical national infrastructure.
Rather than framing the debate as technology versus trade, Ghana would do well to reframe it as a conversation about getting the deployment right. The stakes are high, the intent is sound, and the course correction, if taken now, could position Ghana as a model for AI-driven port reform across Africa.
What Happened at the Ports
Since March 12, 2026, the Ghana Revenue Authority (GRA) has mandated the use of the Publican AI system as the sole basis for customs valuation and import clearance at the country’s ports, including Tema and Takoradi. The government’s justification is hard to argue with. A five-year data review revealed that between 2020 and 2025, commercial banks transmitted over US$127 billion out of the country, yet only US$52 billion worth of physical goods were declared at the border.
The GRA Commissioner-General, Anthony Sarpong, put a figure to the damage, estimating that more than GH¢11 billion has been lost to misclassification, origin fraud, and undervaluation. Publican AI, which benchmarks declared import values against global trade data, was the government’s answer.
The pushback, however, has been swift and organised. A coalition of freight forwarding and trading bodies, including the Ghana Union of Traders’ Associations (GUTA), the Ghana Institute of Freight Forwarders, the Customs Brokers Association of Ghana, and the Freight Forwarders Association of Ghana, launched a strike in mid-April, citing unpredictable duty assessments, prolonged clearance delays, and a near-absent appeals mechanism.
GUTA’s president reported that some duties had “quadrupled”, with assessments leaping from “GHC50,000 to GHC400,000” on the same goods, according to Asaase Radio. Trade experts have also flagged a troubling asymmetry: where the AI’s valuation is “higher than a trader’s declared value, it is accepted; where it is lower, it is ignored”.
IMANI Africa also argues that this puts the system in direct conflict with the WTO’s Customs Valuation Agreement and Sections 67 and 68 of Ghana’s own Customs Act, both of which require that transaction value serve as the primary basis for duty assessment.
The Global Lesson Ghana Is Missing
What makes Ghana’s situation instructive is not that it tried to use AI at its ports; that ambition is well-placed. It is that the country skipped the steps that have made AI work elsewhere. Singapore’s Maritime and Port Authority handles an average of 130,000 vessel calls annually, and its TradeTrust platform has cut veterinary inspection delays for live animal shipments by as much as 63 percent.
The key distinction is that Singapore’s approach embeds AI within a transparent, legally established customs framework, one where overrides follow clear criteria, traders understand the basis for any reassessment, and dispute resolution is accessible. Technology was brought in to strengthen the process, not circumvent it, with human officers remaining central to final decisions.
Rotterdam, Europe’s busiest port, tells a similar story. Its digitalisation of customs processes delivered a reduction in processing time of more than 30 percent and cut the number of trader complaints by as much as 80 percent, results achieved without inflating valuations or triggering trade disruptions.
In China, the Port of Qingdao has built the world’s first air-track intelligent container transport system, integrating AI, 5G, and automated guided vehicles across its terminal operations. Its sea-rail intermodal system alone saves approximately US$57 per container and reduces waiting time by 24 hours.
Even within Africa, peers are moving more carefully. Kenya’s Port of Mombasa is piloting AI-enabled scanners that analyse cargo images in seconds, flagging suspicious consignments for targeted inspection while allowing genuine cargo to move faster. Egypt, Morocco, Nigeria, and South Africa are similarly deploying AI as an intelligence layer on top of existing customs processes, rather than as a mandatory override of declared values.
A Smarter Way Forward
The core problem with Ghana’s rollout is not the technology itself; it is the sequence. A system with this much power over trade and livelihoods required a period of parallel running, where Publican AI’s outputs were tested and compared against existing valuations without a binding effect. It required algorithm transparency so that traders and their agents could understand and challenge assessments. And it required a decentralised appeals infrastructure, not a committee that convenes twice a week in Accra, forcing importers from Takoradi and Paga to travel hundreds of kilometres to dispute a machine’s figure.
Retraining the algorithm on Ghana-specific trade data, rather than global averages that do not reflect the realities of West African commerce, is not optional. It is foundational.
The Ghanaian government has a legitimate and pressing case for modernising its customs architecture. The revenue leakages it has identified are real, and digital tools are the right solution. But as the global evidence makes clear, the countries that have succeeded did not choose between revenue protection and trade facilitation; they pursued both at the same time, through technology that was transparent, legally grounded, and built with the trade community rather than imposed upon it. Ghana still has the chance to get this right.
