THE Department of Communications and Digital Technologies has published a sweeping draft ‘National Artificial Intelligence Policy‘ that aims to put South Africa at the forefront of the global AI revolution. But nothing could be further from the truth.

Buried in the fine print is a proposal that should make every motorists, data users and everyday consumers sit up and take notice: a new AI Insurance Superfund modelled directly on the notoriously troubled Road Accident Fund (RAF).

Inside 86 pages of technocratic govspeak alongside big-ticket ambitions: smarter hospitals, AI-powered schools in every language, job-creating startups, and even digital preservation of indigenous arts and music, you will find a proposal to once again, milk consumers, as cabinet views AI not as a windfall, but rather a new cash-cow, an unlimited source of revenue.

Public comments close on 10 June 2026 – giving ordinary South Africans just eight weeks to have their say before the policy is finalised.

The Upside: Benefits for Ordinary People

On paper, the policy is ambitious and people-focused. It wants AI to:

• Cut costs and create jobs in healthcare, education, agriculture and public services
• Teach AI skills from primary school through university, with community hubs in townships and rural areas
• Translate government services and education into all 12 official languages in real time
• Protect cultural heritage by digitising indigenous languages, stories and music
• Bridge the digital divide so marginalised communities aren’t left behind

If delivered well, the policy could mean faster hospital diagnoses, cheaper farming advice via apps, and more opportunities for young entrepreneurs. It aligns with global trends like the EU’s risk-based AI rules and UNESCO’s ethics guidelines, while stressing South Africa’s constitutional values of fairness and redress.

The Downside: Bureaucracy, Costs and a Potential “AI Tax”

Critics – and a close reading of the document – warn the plan risks creating exactly the kind of bloated, expensive government machinery South Africa can ill afford.

The draft calls for at least seven new or massively expanded institutions: a National AI Commission, an AI Ethics Board, an AI Regulatory Authority, an AI Ombudsperson, a National AI Safety Institute, an Integrated AI Monitoring Centre, and the controversial AI Insurance Superfund. Add in a new cross-regulator forum involving ICASA, the Information Regulator and others, and you have a recipe for overlapping mandates, duplicated salaries and endless meetings.

All of this must be funded somehow – at a time when the national budget is already stretched. The policy talks of “securing funding” in Year 2 (2026/27) and incentives like tax breaks for big players, but offers no detailed costings or budget line.

The Most Controversial Idea: An AI Version of the Road Accident Fund

The biggest red flag for consumers is the proposed AI Insurance Superfund. It would act as a no-fault compensation scheme for anyone harmed by AI decisions – think biased loan rejections, wrong medical diagnoses by AI, or autonomous systems gone wrong – when it’s “difficult to determine” who is liable.

The model? The Road Accident Fund.

That should ring alarm bells. As recent parliamentary inquiries and court rulings have shown, the RAF is drowning in an estimated R400–R500 billion in liabilities, with assets of just R33 billion. It faces adverse audit opinions, massive backlogs (over 440,000 unpaid claims), fraud scandals involving law firms, and repeated corruption allegations. Even after hiking the fuel levy to R2.25 per litre, it still can’t pay its bills. Victims wait years, while the fund remains a “significant fiscal risk” to the national budget.

If the AI Superfund follows the same template – and the draft explicitly says it is “modelled after” the RAF – consumers could ultimately foot the bill through a new levy on AI services, data usage, cloud computing or even everyday apps and devices. That’s effectively an AI tax passed straight to your pocket, even if the policy never uses those exact words.

International experience offers a warning: the European Union’s AI Act deliberately rejected a mandatory public compensation fund or compulsory insurance scheme, preferring to rely on existing private liability laws and targeted regulation. Other countries are developing private AI liability insurance products rather than creating another state-run mega-fund.

What This Means for You – and Why You Should Act Now

If the policy sails through unchanged, South Africans could face:

• Higher prices for AI-powered services (banking apps, insurance quotes, delivery drones, medical tools)
• Slower innovation if red tape and compliance costs scare off startups
• Yet another layer of government bureaucracy that duplicates work already done by the Information Regulator (privacy) or SAHRC (human rights)
• Risk that good intentions around ethics and inclusion get lost in administrative chaos

But the policy is still a draft. The 60-day public comment window is your chance to shape it.

Deadline: 10 June 2026, 16:00
Email written submissions to aipolicy@dcdt.gov.za with the subject line “Draft South Africa National Artificial Intelligence (AI) Policy”. Mark clearly if you want your comments kept confidential.

You don’t need to be a tech expert. Simple, clear points carry weight: “I support skills training but oppose creating another RAF-style fund that could raise my costs”; “Please cost the new institutions before approving them”; “Focus on private-sector innovation and existing regulators instead of new bureaucracies.”

The government says this policy will define AI’s role in South Africa for the next decade. Whether it delivers genuine transformation or becomes another expensive state experiment depends, in part, on what consumers say in the next eight weeks.

Your phone, your bank app, your child’s future classroom – they will all run on AI. Now is the time to make sure the rules protect you, not burden you. Speak up before 10 June.

[Disclaimer: This piece is written by AI with a human-in-the-loop ]