With the recent AI policy scandal in the Department of Digital Communications and Technology, South Africa should ask whether it needs a restrictive artificial intelligence policy framework in the first place. Emulating Europe’s regulatory straitjacket would stifle our low-growth economy and impose unaffordable compliance burdens on start-ups. America’s light-touch approach proves the better path.
Communications minister Solly Malatsi recently found himself in hot water. His department released a draft National Artificial Intelligence Policy Framework, only to withdraw it after it emerged that the document was riddled with fictitious references – classic artificial intelligence “hallucinations”. A policy meant to govern artificial intelligence (AI) had itself been undermined by the very technology it sought to tame.
Malatsi rightly pulled the document, but the episode raises a key question: Why does South Africa need an AI policy at all?
China
The instinct to regulate AI, like many other forms of regulation, is intuitive.
Policymakers in the West and South Africa speak of ensuring the technology delivers only ideal, beneficial outcomes: “safe” AI that aligns with social values, protects privacy and equality, and safeguards against abuse. This noble posture on paper ignores the geopolitical reality. The People’s Republic of China will never play by these rules.
China is laser-focused on building the most powerful AI systems possible for every domain: state surveillance, military applications, commercial dominance, scientific research, and cultural influence. Beijing has no interest in European-style risk classifications or South African socio-economic goal alignment that could slow deployment.
While regulators in Brussels and Pretoria debate consent forms and ethics boards, Chinese labs will iterate faster, train on more data, and deploy at scale. The result is predictable: regulated jurisdictions will fall behind their key strategic adversary.
This is not abstract pondering, but something we already see play out in practice.
The United States has thus far maintained a dynamic, relatively light-touch environment that has allowed companies like OpenAI, Anthropic, Google, and xAI to lead the world.
Europe, by contrast, has the heavily prescriptive Artificial Intelligence Act, Digital Services Act, and General Data Protection Regulation (GDPR) framework. The outcome is a sclerotic European AI sector that lags significantly, with innovation fleeing to less burdened shores.
Brussels effect
As our recent report, “The Brussels effect in South African public policy” (which I co-authored with colleagues at the Free Market Foundation and EPICENTER), elaborates: South Africa habitually imports these European models.
South Africa’s Protection of Personal Information Act (POPIA) mirrors the GDPR. The Competition Commission’s market inquiries into online platforms draw explicit inspiration from the Digital Markets Act and Copyright Directive. The withdrawn AI Policy Framework itself referenced alignment with “global” (read: European) governance standards.
This is the Brussels effect in action: not merely commercial compliance by firms, but policymakers in Pretoria treating European regulations as international best practice, despite our vastly different economic context.
South Africa’s economy is low-growth, high-unemployment, and structurally challenged. We have no internationally competitive AI industry to “protect” through heavy regulation. Imposing compliance burdens designed for wealthy, mature economies on our start-ups, researchers, and SMEs will not create safety but entrench stagnation.
Data restrictions like those in POPIA already hinder AI training and research. Lay on top European-style tiered risk assessments, mandatory conformity checks, and ideological requirements around “equality” and “sustainability”, and we guarantee failure. Our National AI Policy Framework even suggested developing AI in line with South Africa’s “socio-economic goals and values” – a phrase that, in domestic governance, often signals rent-seeking and bureaucratic control rather than open innovation.
Some anti-AI activists might get a kick out of this. “Oh no, the AI companies can’t STEAL data and train on the work of real authors and artists? Good!” they might screech. And while this might feel good in the moment, development in China carries on apace, with their AI models – certainly those being developed privately inside state institutions – training on any and all data they can get their hands on.
There is no “national sovereignty” when it comes to AI. As an almost exclusively-online phenomenon, where data truly knows no border, there are simply AI iterations that succeed and iterations that fail, with the latter category tending to find itself in jurisdictions believing with some measure of hubris that they can rein the technology in.
Regulation as a shortcut to safety is an illusion when your competitor refuses to play by the same rules. The United States succeeds because it bets on human ingenuity and market discovery. Europe regulates itself into irrelevance. South Africa should not imitate the latter.
A better approach
Instead, we should embrace a different spirit. As Marc Andreessen writes in his Techno-Optimist Manifesto:
“We believe Artificial Intelligence is our alchemy, our Philosopher’s Stone – we are literally making sand think.”
“Artificial Intelligence is best thought of as a universal problem solver. And we have a lot of problems to solve.”
“We believe any deceleration of AI will cost lives. Deaths that were preventable by the AI that was prevented from existing is a form of murder.”
Andreessen’s vision is one of abundance and progress – phenomena that do sometimes involve some measure of risk-taking.
Intelligence, both human and artificial, is the ultimate engine of human flourishing. Technology is the lever that lets us do more with less. South Africa’s existential crises, unemployment, energy shortages, poor service delivery, violent crime, are precisely the problems AI is suited to attack when in the hands of commerce and civil society.
Rather than another policy paper chasing European approval, South Africa needs a pro-innovation agenda: light-touch rules that encourage experimentation without having to hire compliance officers, exempt research from onerous consent requirements, protect open data flows where possible, and reject censorship-adjacent “safety” measures. Draw inspiration from the dynamic American model, not Brussels.
Resist the European Union’s imperialism of expedience.
The Malatsi episode was a humiliating reminder of what happens when bureaucracy tries to box in a fast-moving technology it barely understands. Let us hope it is also a turning point.
South Africa does not need an AI policy to manage risks that might never materialise, at the expense of opportunities we desperately need. We just need to let the market be.
[Image: https://www.pexels.com/photo/webpage-of-ai-chatbot-a-prototype-ai-smith-open-chatbot-is-seen-on-the-website-of-openai-on-a-apple-smartphone-examples-capabilities-and-limitations-are-shown-16380906/]
The views of the writer are not necessarily the views of the Daily Friend or the IRR.
If you like what you have just read, support the Daily Friend