Ghana Presents Publican AI Customs Gains At BRITACOF

Ghana's revenue authority has presented its use of artificial intelligence in customs and tax administration to an international forum, reporting that the Publican AI system generated an additional GH¢1 billion in customs revenue in April 2026.

The figure was not a new October revenue result. It was an earlier outcome highlighted during Ghana's presentation at the seventh Belt and Road Initiative Tax Administration Cooperation Forum, held in Jakarta from 13 to 15 October. Preserving that timeline is essential to avoid turning a conference presentation into a fresh monthly performance claim.

Ghana News Agency reported that Elsie Appau-Klu, technical adviser to the commissioner-general and leader of the Ghana Revenue Authority delegation, described how digital services and risk tools are being used to improve compliance and identify revenue leakage.

Publican AI analyses information from importers, exporters, trade documents, bank transfers and historical customs records. It is designed to identify possible undervaluation, misdescription, over-invoicing and other inconsistencies for closer examination. GRA officials have consistently described it as decision support rather than an autonomous customs officer.

That difference should remain visible. The authority told the forum that final decisions remain with human officials and identified privacy, legal authority, data security, oversight and taxpayers' rights to object and appeal as necessary safeguards. These protections determine whether automated risk assessment improves administration or merely makes contested decisions faster.

The system was piloted in February and fully deployed on 12 March. In a June explanation, customs officials said it could reduce document-review time and expose inconsistencies between invoices, shipping records and historical values. That can help address genuine under-declaration and make similar consignments more consistent.

But implementation has also produced resistance. Agricultural-input businesses and freight forwarders previously complained that some valuations were excessive, delayed clearance and increased demurrage costs. CropLife Ghana warned that disputed assessments could affect the cost and availability of farm inputs.

Those accounts do not necessarily invalidate the revenue result. They show why collection gains cannot be the only measure of success. Ghana should publish how many declarations are flagged, how often officers accept or overturn the recommendation, the time required to resolve appeals and whether comparable goods receive more consistent treatment.

The authority should also distinguish additional assessments from cash actually collected. A system may identify a higher customs value, but disputes, appeals and delayed clearance affect what reaches the treasury and at what economic cost. Reporting both figures would make the performance claim easier to evaluate.

The GRA's decision to retrain auditors as data analysts and risk managers is a constructive signal. Public institutions often describe AI as a route to staff reduction, even when the more immediate need is better judgement. Officers who understand both customs rules and model limitations are more capable of identifying bad data, unusual cases and systematic bias.

Publican AI is one of Africa's most consequential public-sector AI deployments because it affects state revenue and the daily cost of cross-border trade. Ghana's presentation at BRITACOF gives the project international visibility. Sustained legitimacy, however, will depend on transparent performance data and an appeal process that works as quickly as the automated system itself.

Sources: Ghana News Agency on BRITACOF, GNA's system explainer and GNA on importer concerns.