As autonomous vehicles (AVs) evolve from concept to real-world use, South Africa faces pressing legal questions: if an AI-driven vehicle causes harm, who bears liability? Under current South African law, liability for road accidents is largely governed by the Road Accident Fund Act 56 of 1996 (the RAF Act).
The RAF is a statutory insurance fund that provides compulsory compensation to anyone injured in a motor vehicle accident on South African roads. Its object is to pay compensation for “loss or damage wrongfully caused by the driving of motor vehicles.” This includes indemnity for those who cause accidents and benefits to injured victims or their dependents.
In practice, a claimant must show two key elements for a valid RAF claim:
- Bodily injury “caused by or arising from the driving of a motor vehicle.”
- That the injury was due to the negligence or other unlawful act of the driver, owner, or their servant.
The RAF thus functions as a form of social insurance. It protects victims by providing compensation without requiring litigation directly against an insured driver. Courts have emphasised that the RAF’s object is to afford compensation when someone is injured as a result of negligent driving, including cases where the negligent driver cannot be identified.
However, the statute’s current language assumes a human driver whose negligent conduct causes harm – a framing that does not neatly translate to autonomous vehicles controlled by algorithms.
The RAF Act defines a “motor vehicle” as any vehicle designed or adapted for propulsion or haulage on a road by means of fuel, gas, or electricity. This definition is broad and would include autonomous vehicles as long as they meet these physical criteria. In previous cases, courts have interpreted what qualifies as a motor vehicle based on design and purpose.
Thus, an AV involved in an accident would likely fall within the RAF Act’s definition, making injuries caused by it prima facie compensable under the RAF, provided the remaining elements are satisfied. The difficulty arises with fault, specifically, how to attribute negligence when no human is actively driving at the moment of a crash.
Under the statute, a claimant must typically prove the negligent driving of a person, such as a human driver. But with automated systems, the traditional notion of driver negligence becomes ambiguous. In a crash involving an autonomous system, a claimant might need to show that some wrongful act occurred in connection with the vehicle’s operation. For example, poor maintenance, defective design, inadequate software updates, or failure to respond to foreseeable hazards.
However, under the current RAF framework, liability is still tied to conduct that can be attributed to a person or entity. This means that owners, operators, or service providers associated with the vehicle may be implicated in claims, even when a human driver is not at fault. For property damage (like vehicle damage), the RAF does not provide coverage; compensation for damage to vehicles is normally pursued in civil claims against at-fault parties or their insurers.
Current South African legislation does not explicitly address autonomous technology. The RAF Act contemplates human conduct in driving and assigns liability based on negligence in that context. Because AVs can operate without a human driver in control, courts may be asked to interpret the RAF Act to extend to situations where negligence is tied to maintenance, design, or digital control systems rather than human misjudgment. Scholars have proposed reforming the liability regime to explicitly include AVs and clarify how claims should be handled. For instance, requiring AV owners to carry additional private insurance or adapting statutory definitions to include autonomous functions.
In practice, pursuing a RAF claim after an AV accident would still follow the standard procedural requirements: lodging a claim within statutory time limits and proving negligence or wrongful conduct tied to the vehicle’s driving. Because autonomous systems may generate extensive digital logs and sensor data, establishing causation and fault could shift focus toward technical analysis and expert testimony – and potentially lead to parallel civil claims against manufacturers or service providers separate from the RAF claim.
The road ahead for autonomous vehicle liability
South African law currently offers a workable, if imperfect, framework for addressing injuries caused by autonomous vehicles through the Road Accident Fund. While the RAF Act’s scope (compensating victims of “motor vehicle” accidents) likely encompasses AVs, the law’s emphasis on negligent driving by a human driver presents interpretative challenges. In many cases, claimants may need to combine RAF claims with common law actions to ensure full compensation for both personal injury and property damage. As AV technology matures, legislative reform and judicial interpretation will be crucial to ensure the RAF and South African liability law remain effective and fair in adjudicating AI-driven harm.
While every reasonable effort is taken to ensure the accuracy and soundness of the contents of this publication, neither the writers of articles nor the publisher will bear any responsibility for the consequences of any actions based on information or recommendations contained herein. Our material is for informational purposes.