Orodata Science, a Nigerian data science and research organisation, has urged civil society organisations (CSOs) to participate actively in shaping the governance of frontier Artificial Intelligence (AI).
Orodata made the call in a policy brief titled “Pacing Frontier AI: Protecting Rights and Public Interest Access in Nigeria”, released in Lagos on Wednesday.
The organisation said the global debate on slowing the development and deployment of increasingly capable AI systems presents both risks and opportunities for Nigeria and other Global South countries, the News Agency of Nigeria (NAN) reports.
It said the debate followed recent calls by Dario Amodei, Chief Executive Officer of Anthropic, for deliberate pacing of frontier AI development. According to the brief, Amodei proposed stronger independent evaluation, binding technical safety standards and international coordination before AI systems reach potentially dangerous capability thresholds. It added that the proposals received public support from OpenAI Chief Executive Officer Sam Altman and Elon Musk, founder of xAI and Chief Executive Officer of Tesla.
Orodata, however, cautioned that voluntary commitments by technology companies should not be treated as legally binding agreements or as substitutes for enforceable regulatory frameworks. It said Nigeria must guard against the systemic risks of unchecked frontier AI while also preventing digital marginalisation and excessive concentration of technological power.
The organisation urged Nigerian CSOs to make use of existing domestic and continental frameworks, including the Nigeria Data Protection Act 2023, the National Artificial Intelligence Strategy and the African Union Continental AI Strategy. It said these should be used to demand verifiable commitments from AI developers, funded African red teaming, data sovereignty protections and equitable access to AI technologies.
Orodata explained that “pacing” does not mean an abrupt halt to software development or the withdrawal of existing digital services. Rather, it involves introducing regulatory checkpoints, red teaming phases and capability thresholds before more advanced AI systems are trained or deployed.
It identified independent evaluation, harmonised industry safety standards and multilateral coordination as the key elements of responsible frontier AI governance. It said independent evaluators should be given sufficient access to AI development processes to assess models throughout their training and deployment lifecycle.
The organisation also warned that Nigeria’s dependence on foundational AI models developed and governed largely outside Africa could create strategic vulnerabilities. According to the brief, changes by foreign AI platforms, including pricing, licensing conditions, access restrictions or the withdrawal of application programming interfaces, could disrupt Nigerian startups and public interest initiatives. It said these include healthcare programmes, educational platforms and governance initiatives.
Orodata urged policymakers to consider everyday AI harms alongside potential catastrophic risks. It identified linguistic and cultural marginalisation, political disinformation, deepfake manipulation, financial fraud and the exploitation of African data workers as concerns requiring greater attention.
It said frontier AI systems should be evaluated for performance in Nigerian languages, socio-political contexts, ethnic diversity and local fraud patterns. It also called for greater investment in Nigerian universities, research institutions and CSOs to conduct AI safety testing and red teaming tailored to African realities.
The organisation urged international donors, development finance institutions and technology companies to provide dedicated technical and financial support for such efforts. It further recommended that Nigerian CSOs be included in international AI safety networks and consultations, so that Global South perspectives are represented.
On regulation, Orodata called for tiered liability and proportional compliance requirements that distinguish between frontier AI developers and downstream users. It said developers should bear responsibility for structural design flaws, baseline safety and inadequate testing, while organisations deploying AI should carry responsibilities proportionate to the nature and potential impact of their applications.
It said automated systems used in consequential decisions should retain meaningful human oversight, listing social welfare allocation and legal or community interventions among the areas that require it.
For nonprofit organisations, Orodata recommended five immediate internal controls. These include auditing the AI applications in use, restricting the transmission of sensitive beneficiary information into consumer AI systems and requiring human verification of AI generated outputs. Others are demanding clear supplier terms on data retention and security, and joining civil society working groups to share knowledge, incident reports and procurement resources.
Orodata said Nigeria should not wait for international consensus or new domestic legislation before taking steps to protect its institutions and citizens from AI related risks. It said responsible AI governance should protect democratic rights and social justice while ensuring that African countries are not excluded from the benefits of technological development.
The organisation said Nigerian civil society has an opportunity to influence the emerging global AI governance framework, and urged CSOs to demand stronger data protection, local technical capacity and affordable public interest access to AI.
Orodata stressed that voluntary restraint by foreign technology executives alone cannot guarantee Nigerian digital sovereignty or protect civic space from algorithmic harm.
