Africa’s growing reliance on digital identity systems is raising concerns that errors could lock people out of banking, telecommunications and essential public services, as interconnected systems increase the risk of a single verification failure.
South Africa’s Department of Home Affairs revealed in March 2025 that its digital verification system had suffered failure rates of up to 50% on verification requests, with responses sometimes taking 24 hours and containing errors requiring manual checks.
The system serves government departments, the South African Social Security Agency (SASSA) and financial institutions, illustrating how one unreliable verification layer can affect multiple sectors. Following an upgrade, the department reported an error rate below 1%.
Nigeria experienced a related disruption in July 2024, when the Nigerian Communications Commission ordered telecom operators to reactivate lines blocked following National Identification Number (NIN)-SIM verification problems. The incidents do not establish that errors spread across all connected services, but demonstrate the consequences when identity verification fails at scale.
Allan Dabre, Senior Manager for Technology Compliance and AI Risk at PwC, said interconnected systems could amplify such failures. “One identity unlocks everything,” Dabre said, warning that incorrect records could restrict access to several services.
He noted that correcting an identity database might not resolve errors in separate records held by banks, telecom operators and public agencies. He advocated human review, alternative verification channels and mechanisms for corrections to reach connected systems.
AI adoption makes accountability increasingly important. The Central Bank of Kenya’s 2025 survey found that half of surveyed financial institutions had adopted AI, with electronic know-your-customer processes and customer onboarding among its applications.
The regulator also identified gaps in data quality, governance and risk management.
Arjun Jaggi, an applied AI researcher specializing in identity verification and biometrics, warned that compromised biometric credentials pose challenges because fingerprints and facial characteristics cannot readily be replaced.
He recommended revocable credentials and recovery systems that allow human intervention when automated verification fails.
AI adds another layer of risk when automated verification or fraud-detection systems produce decisions that individuals cannot understand or challenge. Harish Pathak, an AI governance researcher and IEEE Senior Member, said identity decisions must be auditable so individuals can understand and challenge incorrect outcomes.
As digital identity infrastructure expands, governments and service providers must ensure identity failures do not deny people essential services.
Share
