Retailers Face Legal Risk with AI Chatbots Under Australian Consumer Law

Australian retailers are increasingly adopting generative AI chatbots to enhance customer service, but this shift comes with significant legal risks. Lawyers and regulators warn that companies are responsible for the accuracy of information provided by these AI systems, potentially breaching Australian Consumer Law if chatbots mislead customers or provide incorrect advice.

The Rise of Generative AI in Retail

While many Australian retailers still utilize basic, flowchart-driven chatbots, companies like Woolworths, Kmart, and Bunnings are experimenting with generative AI chatbots capable of more natural and creative interactions. These advanced bots, while offering convenience, are prone to unpredictability, even with implemented safeguards.

Recent Incidents and Legal Precedents

Several incidents highlight the potential for legal issues. Woolworths’ chatbot, Olive, faced criticism for rambling about a fictional mother and providing inaccurate pricing information. Bunnings had to add restrictions to its chatbot after it offered potentially illegal electrical repair instructions to a customer without the necessary qualifications.

The legal implications are not limited to Australia. In Canada, an airline lost a tribunal case after its chatbot incorrectly promised a bereavement discount, resulting in damages and fees. In the UK, a small business owner faced threats of legal action when a chatbot offered a customer an unexpectedly large discount.

Consumer Law and Chatbot Liability

According to Matthew McMillan, head of digital economy at Lander & Rogers, retailers cannot avoid responsibility by blaming the AI. “If a chatbot gives incorrect or misleading information, the retailer can be liable for breaches under the Australian Consumer Law. They can’t shift the blame to the chatbot and claim AI acted independently,” he stated. The law focuses on the impact on the consumer, regardless of whether the information came from a human or a machine.

Misleading pricing information or incorrect advice regarding refunds or returns could constitute a breach of consumer law. A chatbot providing offensive content could also lead to discrimination complaints, defamation claims, or privacy breaches.

Areas of Highest Risk

The Australian Competition and Consumer Commission (ACCC) has previously taken enforcement action against companies for misleading claims regarding refunds and guarantees, including Valve, Sony, Booktopia, Mazda, and The Good Guys. The highest risk area for chatbots lies in handling queries related to refunds and returns, where inaccurate information could easily lead to a breach of consumer rights.

A spokesperson for the ACCC emphasized that retailers will be held accountable for chatbot-provided information and advised customers who believe they have been misled to contact the company or a consumer protection agency.

Retailer Responses and Disclaimers

Bunnings reports that its AI chatbot assists customers with project planning, product information, stock availability, and order tracking, with continuous monitoring of customer feedback. Woolworths states that customers are informed their chat with Olive may contain errors and that the system operates within controlled parameters.

However, McMillan cautions that disclaimers stating the chatbot may be inaccurate are unlikely to protect a company if a consumer is misled. The key factor is whether a reasonable consumer would likely be misled by the information provided.

Current Chatbot Behavior

Recent testing of chatbots at Myer, David Jones, The Iconic, JB Hi-Fi, and other companies revealed that many either avoided answering questions about pricing and returns or transferred the chat to a human agent. The Bunnings bot provided detailed responses, drawing pricing information from live listings and carefully referencing consumer guarantees. Kmart’s bot appeared to be malfunctioning.

Key Takeaways

• Retailers are legally responsible for the information provided by their AI chatbots.
• Misleading or inaccurate information can lead to breaches of Australian Consumer Law.
• Disclaimers may not fully protect retailers from liability.
• The highest risk areas are refunds, returns, and potentially illegal advice.
• Continuous monitoring and accurate information are crucial for responsible chatbot implementation.