With the growing use of Artificial Intelligence (AI) and its integration into corporate systems, a key question arises: can a company be held responsible for misleading statements made by its AI chatbot to a customer?
This issue arose in Moffatt vs Air Canada, a case concerning a refund for a bereavement fare. Although a Canadian small-claims tribunal decided the case in 2024, its ruling does not bind South African courts. Nevertheless, Moffatt offers valuable guidance. As AI-driven customer service tools become commonplace, its reasoning on when a company must answer for its chatbot’s statements provides useful lessons for South Africa. The case also raises important questions about AI chatbots and corporate liability.
Unpacking the Moffatt case
Following his grandmother’s death in November 2022, Mr Moffatt used Air Canada’s website chatbot to research bereavement fares for an urgent flight. The chatbot advised him that he could claim a reduced bereavement rate retroactively. He had to submit a refund application within 90 days of the ticket’s issue date.
Relying on this information, Mr Moffatt booked one-way flights between Vancouver and Toronto at standard rates. He subsequently applied for partial refunds. However, Air Canada rejected his refund request. The airline referred him to a different webpage on its website, which stated that customers could not apply bereavement fares retroactively.
The Tribunal found Air Canada liable for negligent misrepresentation. It applied the principle that Mr Moffatt should occupy the position he would have been in had the misrepresentation not occurred. Accordingly, it awarded him CAD 650.88 in damages, which fell short of his claimed amount of CAD 880.36.
Lessons from the judgment
The central issue was whether Air Canada could be held liable for inaccurate information generated by a chatbot on its website. The case provides an important example of AI chatbots and corporate liability, particularly when businesses use automated systems to communicate with customers.
The Tribunal defined negligent misrepresentation as occurring “when a seller does not exercise reasonable care to ensure its representations are accurate and not misleading”.
Air Canada admitted that its chatbot had used “misleading words”. However, the airline argued that it could not be liable for the chatbot’s statements. It suggested that the chatbot effectively constituted a separate legal entity responsible for its own statements.
The Tribunal rejected this argument and described the submission as “remarkable”. It found that the chatbot formed part of Air Canada’s website. Furthermore, the Tribunal held that consumers could not reasonably be expected to know which section of the airline’s website contained accurate information when contradictory statements appeared on the same platform.
As a service provider, Air Canada owed Mr Moffatt a duty of care. The airline breached the applicable standard because it failed to take reasonable steps to ensure that its website and chatbot provided accurate information. The judgment therefore challenges the assumption that businesses can separate themselves from statements generated by their automated systems. Instead, it demonstrates why organisations must consider AI chatbots and corporate liability when designing, deploying and monitoring customer-facing technologies.
South African perspective
South African law already recognises negligent misrepresentation as a basis for liability under the law of delict. In addition, the Consumer Protection Act 68 of 2008 prohibits misleading representations to consumers. However, no reported South African case has yet tested these principles specifically in the context of an AI chatbot.
Nevertheless, South African courts could plausibly find favour with the reasoning in Moffatt. A company cannot simply treat its chatbot as an autonomous actor detached from its business operations.
If a chatbot misleads a client about the terms, procedures or exclusions of a product or service, the company that deployed it may bear liability. The outcome would depend on the applicable legal principles and the specific circumstances of the case.
This issue matters as chatbots become standard features of client-facing services in South Africa. Although these tools offer genuine efficiency gains, they also create legal exposure. A single misrepresentation can trigger financial, reputational and regulatory consequences. Moreover, this risk may extend to AI systems that third-party service providers use on behalf of companies.
South African courts may also adopt reasoning similar to that of the Canadian Tribunal. Consumers cannot reasonably be expected to distinguish between “official” and “automated” sources of information when a company presents both through the same platform.
In conclusion
As digital adoption accelerates across South Africa, companies should implement governance frameworks, compliance checks and human oversight of AI-generated outputs. They should act now rather than wait for the first local test case.
Ultimately, businesses must recognise that AI chatbots and corporate liability are increasingly interconnected. Organisations should establish clear accountability for chatbot outputs, regularly review automated responses and ensure that their digital platforms provide consistent, accurate information. These measures can help businesses manage legal risks while maintaining consumer trust.
 |
 |
 |
| Erwyn Durman | Partner | mail me | | Manno Manavhela | Associate | mail me | | Precious Maphupha | Candidate Attorney | mail me | |
| | Webber Wentzel | | ||
