AI in the tax office, the newsroom, and the fine print
This edition covers a wide spread of AI activity touching South African daily life: from SARS using AI to identify taxpayers who may not be compliant, to a fintech building AI tools for small merchants, to new data showing South Africans are turning to AI for their news. Running alongside these local developments are two stories that set the broader context: a US government order that cut off access to Anthropic's AI models for foreign users without warning, and an analysis of how African countries, South Africa included, are engaging with AI through skills programmes and symbolic partnerships while the infrastructure and ownership that generate lasting value remain elsewhere. A piece on AI governance standards and a column on why AI systems often cannot explain their own decisions round out an edition that spans government, business, labour, and the information environment.
Policy & governance
AI is now hunting tax cheats in South Africa
TechCentralPolicy
South Africa's Revenue Service (SARS) is using artificial intelligence (AI) – software that finds patterns in large amounts of data – to identify taxpayers who may not be meeting their obligations, according to TechCentral. The system also supports customer service functions, though SARS says human officials retain the final say on any enforcement decision. The development matters for South Africans because it signals a significant shift in how the country's main tax authority monitors compliance, with implications for both individual taxpayers and businesses.Communications Minister to deliver keynote on AI at this month’s Social Media Summit
Hypertext (htxt)· SponsoredPolicy
South Africa's Minister of Communications and Digital Technologies, Solly Malatsi, is set to deliver the keynote address at the fourth annual Social Media Summit for Government in Johannesburg on 30 June, according to Hypertext. The appearance comes shortly after his department faced sharp criticism over a Draft National AI Policy that was found to contain AI-generated text (content produced automatically by an AI system rather than written by officials) that could not be properly cited or verified. Organisers say his address will cover the government's vision for digital transformation and rebuilding trust between the state and citizens, making the summit a notable moment to watch how the department accounts for the policy misstep.Anthropic got hit by export rules nobody understands
The Verge — AIPolicy
US AI company Anthropic was ordered by the Trump administration this week to cut off access to two of its newest AI models for all foreign nationals, including users based inside the United States and its own staff, citing national security grounds. The Verge reports this appears to be the first time US export controls, rules that restrict the transfer of sensitive technology to other countries, have been applied directly to an AI model in this way. For South African users, researchers, and businesses that rely on Anthropic's tools, the episode signals that access to leading AI systems can be suspended without warning by US government order, a risk that applies to any country outside the United States.Also reported by Ars Technica — AI, The Verge — AI, The Verge — AI, TechCentral, The Verge — AI
What Ruto’s meeting with Sam Altman tells us about Kenya’s AI ambitions
TechCabalPolicy
TechCabal analyses Kenya's announcement that President William Ruto met OpenAI chief executive Sam Altman at the G7 Summit to discuss establishing an OpenAI Academy in Nairobi, noting that the statement contained no investment figures, timelines, or firm commitments. The piece argues that Africa's engagement with AI is increasingly built around skills programmes and symbolic partnerships rather than the compute infrastructure, capital, and intellectual property ownership that determine who captures lasting value from the technology. South Africa is named alongside Kenya and Nigeria as sharing the same structural tension: a real pool of software talent, but limited domestic ownership of the infrastructure and research that shape where AI's economic gains ultimately flow.ISO 42001 helps organisations prepare for the realities of AI governance
ITWeb· SponsoredPolicy
ITWeb reports on ISO 42001, an international standard that sets out a management framework for how organisations should govern their use of artificial intelligence. The standard provides structured guidance on risk, accountability, and oversight for AI systems, and is relevant to South African companies and public bodies that are building or adopting AI governance policies. As South Africa develops its own regulatory approach to AI, aligning with an internationally recognised standard could help local organisations demonstrate responsible practice to regulators, partners, and customers.
Business & economy
Fintech Yoco thinks South Africa’s small businesses need fewer apps
TechCabal· SponsoredBusiness
South African fintech Yoco used its annual product event in Johannesburg this week to announce more than 20 new features, including a planned AI assistant that will analyse transaction data and business performance to help small merchants make decisions, according to TechCabal. The company acquired Dyner.ai, a software business built for small and medium enterprises, in May, and expects to launch the AI tool in the third quarter of 2026. The announcements reflect a broader push by Yoco, which serves more than 200,000 merchants across South Africa, to move beyond card payments and offer small businesses a single platform covering accounting, inventory, customer loyalty and cash flow management.
Society & work
South Africans turning more to online influencers and AI for news: report
Business Day / BusinessLIVESociety
South Africans are increasingly turning to online influencers and AI tools to get their news, according to the Reuters Institute and Oxford University's Digital News Report 2026, as reported by Business Day. The findings point to a shift in how citizens encounter and consume information, with AI-generated or AI-curated news becoming a distinct source alongside traditional outlets. For South Africa, where public trust in media and access to reliable information are ongoing concerns, the trend raises questions about accuracy, accountability, and who shapes what people know.AI, robots and the future of work
Moneyweb· SponsoredSociety
PSG Wealth chief investment officer Adriaan Pask argues, in Moneyweb, that AI and robotics will reshape the labour market but that roles requiring relationships, oversight, and experience are likely to hold their ground. The piece reflects a growing conversation among South African financial institutions about how workers and investors should position themselves as automation spreads across industries.
Technology & infrastructure
Neural networks and deep learning
Business Day / BusinessLIVETechnology
Writing in Business Day, Rufaro Mafinyani explains why neural networks (software systems loosely modelled on the brain that learn by processing large amounts of data) are effective at recognising patterns yet often cannot account for how they reach a conclusion. The piece, aimed at a South African general audience, addresses a practical concern for local businesses and regulators: when an AI system makes a consequential decision, such as approving a loan or flagging a job application, the inability to explain that decision raises questions of fairness and accountability.