Safaricom sued over AI use, petitioner cite threats to consumer rights and fairness
The lawsuit also challenges the use of algorithmic credit scoring in products such as Fuliza, as well as biometric data collection and AI-powered services in sectors like healthcare and agriculture
A Kenyan public interest litigant has moved to court seeking to block and regulate Safaricom PLC’s growing use of artificial intelligence (AI), Zuri Chatbox in customer-facing services, arguing that the systems operate without sufficient transparency, accountability, or human oversight.
In a petition filed by Victor Odhiambo, he accused Safaricom of deploying AI-driven systems across its operations in ways that allegedly infringe on constitutional rights, including privacy, consumer protection, and fair administrative action as provided under Articles 31, 46, and 47 of the constitution.
In his petition, Odhiambo argues that the Chief Executive Officer (CEO) Peter Ndegwa’s led telecoms giant’s use of AI in areas such as customer service, dispute resolution, mobile money transactions, and credit scoring decisions exposes millions of users to automated decisions without adequate safeguards or accessible human review mechanisms.
The petitioner says that the system has effectively replaced human support in critical situations, leaving consumers stranded during financial emergencies.
“Consumers are not consistently provided with intelligible explanations, prior notice, or meaningful access to human review,” the petition states.
The lawsuit also challenges the use of algorithmic credit scoring in products such as Fuliza, as well as biometric data collection and AI-powered services in sectors like healthcare and agriculture.
The petition also raises concerns over Safaricom’s Zuri AI chatbot, which handles large volumes of customer queries and is integrated into platforms including WhatsApp.
Odhiambo argues that the system limits access to human customer service, particularly when users need urgent assistance or dispute resolution.
He maintains that Safaricom’s expanded reliance on AI is not merely a technical upgrade but a structural shift in service delivery that raises fundamental questions about consumer rights, accountability, and access to justice in automated systems.
The petitioner contends that this creates a risk of discriminatory or unfair outcomes.
He is seeking a court declaration that consumers affected by automated decision-making are entitled to clear notifications, meaningful explanations of decisions, and timely human intervention where adverse outcomes arise.
Among his prayers to the court, Odhiambo wants Safaricom compelled to file a compliance report detailing all AI systems used in services affecting consumers, including those related to financial services, customer support, and health-related tools.
He also wants the company ordered to publish transparency notices outlining how its automated systems operate, the data they use, and available complaint and redress mechanisms.
The petitioner further seeks orders requiring Safaricom to establish human review processes for issues such as fraud alerts, mistaken transactions, and account restrictions, ensuring customers can escalate disputes to human agents.
He is also asking the court to compel the company to disclose the criteria used in credit scoring systems such as Fuliza and to conduct fairness audits to ensure compliance with anti-discrimination provisions under Article 27(4) of the Constitution.
Additionally, Odhiambo wants Safaricom barred from introducing mandatory biometric enrolment requirements unless they are legally justified and alternative options are provided.
Odhiambo is also urging the court to direct the Communications Authority (CA) of Kenya and the Competition Authority of Kenya (CAK) to monitor AI compliance with consumer protection and fairness standards.
In his filing, Odhiambo highlights Safaricom’s evolution into a major AI-driven technology platform, noting its reported $500 million investment in artificial intelligence systems.
He argues that these technologies now influence services used by more than 46 million customers across mobile money, credit, and customer care platforms.
He specifically cites the M-PESA Fintech 2.0 system, which reportedly processes over 100 million transactions daily using machine learning models, alleging that legitimate transactions are sometimes flagged as suspicious without clear explanations or timely human review, potentially violating the Data Protection Act, 2019 and constitutional guarantees of fair administrative action.
