Inconsistent responses from leading artificial intelligence chatbots on politically sensitive topics threaten to erode public trust and complicate the emerging AI governance landscape across Africa.
The warning from African AI governance experts follows a Meta Oversight Board study, which revealed that major AI models are significantly more likely to refuse requests critical of governments in countries with restrictive speech laws than in open democracies.
Risking an extension of government restrictions on free expression, AI systems require developers to improve transparency and conduct stronger human rights assessments, warned the Board.
Africa faces a structural disadvantage because most large language models are trained predominantly on English-language content from the US and Europe, says Faith Mpara, technology policy specialist at Omnisphere Partners.
“Political content about Cameroon's Anglophone crisis, Ethiopia's Tigray conflict or Nigeria's #EndSARS protests exists in far smaller volumes online than comparable Western events,” says Mpara. “Users need to treat these technologies as a source, not the full picture.”
Global AI safety policies are often built around Western interpretations of harmful content, making it easier for legitimate civic discourse in Africa to be incorrectly flagged as incitement or disinformation, adds Mpara.
“Who defines what constitutes 'disinformation' or 'incitement' matters because enforcement can easily drift towards censorship,” says Mpara.
Attributing inconsistencies to government pressure alone is inaccurate, as AI models are shaped by a combination of safety policies, training data and regulatory risk assessments, cautions Ahmed Inuwa Sanni, founder of Locally Led Development.
“There isn't a single cause,” says Inuwa Sanni. “AI models are designed to reduce harm, so they often apply different safety measures to politically sensitive topics. Africa also faces a challenge because there is far less high-quality, publicly available digital content about many African countries, languages and political contexts. That means AI systems may rely on incomplete, outdated or less representative information.”
Improving African data representation, supporting more local languages and increasing transparency around how AI systems make decisions will be critical to building public trust as chatbots become a common source of political information, adds Inuwa Sanni.
Rather than regulating opinions, policymakers should focus on transparency, accountability and independent oversight while protecting freedom of expression, says Inuwa Sanni.
Uneven AI governance across African markets leaves technology companies making policy decisions in practice, says Testimony Akinkunmi, compliance lead at OnPoint.
“While Nigeria's National Assembly is still dragging its feet on AI legislation, the Central Bank isn't waiting around,” says Akinkunmi, referring to the regulator's AI risk management guidance. “The actual AI governance framework running most African countries is often a terms-of-service agreement drafted in California or Stockholm.”
Companies also factor political risk into their moderation decisions, points out Akinkunmi, citing Nigeria's seven-month Twitter ban in 2021 as evidence that platforms remain conscious of regulatory retaliation when designing safeguards for politically sensitive content.
Ensuring that transparency, accountability and local context evolve as quickly as the technology itself remains Africa's key challenge as AI becomes a central source of news and civic information.
Share
