In 2018 Zimbabwe signed China's first Belt-and-Road AI export deal with CloudWalk to build a national facial-recognition database for airport, rail and bus-station security, letting CloudWalk train its algorithms on citizens' biometric data with no published oversight.
Details
Maturity
Established
Promoter
Government of Zimbabwe; CloudWalk Technology
Period
Agreement signed April 2018; expanded through the 2020s
Keywords
facial recognition, surveillance, public security, biometric data, Belt and Road Initiative
Context
On 13 April 2018, Zimbabwe's government signed a strategic cooperation agreement with Guangzhou-based CloudWalk Technology — reported as China's first Belt and Road AI-technology export to Africa — to build a national facial-recognition database for public security use, following a comprehensive strategic partnership recently established between Zimbabwe and China.
Objectives
Named applications include a "smart" financial-services network and "intelligent security" screening at airports, railway stations and bus stations, aimed at curbing crime; CloudWalk's 3D-light facial-recognition software was marketed as more accurate on dark-skinned faces than comparable Western systems.
Activities
The arrangement explicitly permits CloudWalk to train its commercial facial-recognition algorithms on Zimbabwean citizens' biometric data — reporting by Quartz Africa describes this as building "one of the world's most... racially-diversified facial recognition databases" for a company that also supplies technology behind China's domestic "Police Cloud" surveillance system. By 2023, Africa Defense Forum reported the programme had expanded further into the country's broader surveillance infrastructure.
Conclusions
The deal has drawn sustained independent scrutiny. Digital-rights researcher Arthur Gwagwa told Coda Story that such deals let "Beijing use developing countries as laboratories to improve its surveillance technologies" while Chinese firms "get to train their algorithms on darker-complexioned Africans," warning that Zimbabwean pro-democracy activists could be misclassified and targeted through the same infrastructure. No source discloses the system's accuracy, the scale of data collected, an independent audit, or any public legal safeguard governing its use — CloudWalk declined interview requests and told the Global Times only that the government sought "a comprehensive package plan." This is a documented surveillance partnership with unresolved oversight and human-rights concerns, not a demonstrated public-value success.
Implementation
Indicative cost
High (€500k–€5M)
Time to results
Long (> 3 years)
Staffing & skills
CloudWalk Technology (Guangzhou) providing facial-recognition algorithms and biometric database, Government of Zimbabwe security agencies deploying screening at airports, railway and bus stations
Conditions for success
No published independent audit, accuracy disclosure, or legal safeguard currently governs use of citizens' biometric data
Meaningful independent oversight and a public consent/legal framework would be needed before the deal could be assessed as a responsible public-security practice
CloudWalk has declined public interview requests on system accuracy or safeguards, limiting external verification
Common failure modes
Digital-rights researchers (Arthur Gwagwa, quoted by Coda Story) warn the arrangement lets algorithms be trained on Zimbabwean citizens' biometric data without disclosed consent, with risk of misclassification and targeting of pro-democracy activists via the same infrastructure.
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.