Companies Face Scrutiny and Liability for AI Chatbot-Provided Information

Australian Retailers Face Legal Liability for AI Chatbot Misinformation

Recent incidents involving artificial intelligence (AI) chatbots generating misleading or incorrect information have prompted discussions about company liability, consumer protection, and the urgent need for greater transparency. As more Australian retailers integrate generative AI into customer service, legal experts and regulators affirm that companies are responsible for the information their chatbots disseminate, regardless of disclaimers.

Incidents of Alleged Deception and Misinformation

Several events have highlighted growing concerns about the accuracy and disclosure of AI chatbots:

A journalist inquiring with Aavegotchi, a company reportedly based in Singapore, received a rapid, detailed response attributed to "Alex Rivera, Community Liaison." When questioned if the response was AI-generated, a second rapid reply denied AI involvement, stating it was from a human member and signed "Alex (real human)." Subsequent attempts by the journalist to contact "Alex Rivera" via provided contact details, including a phone number and a manager's email, were unsuccessful.

This sequence of events, and others, raises questions regarding "AI hallucinations."

"AI hallucinations" describe instances where AI systems generate information that appears accurate but is factually incorrect or misleading.

• Bunnings Incident: A Bunnings chatbot reportedly provided electrical advice that, by law, requires a licensed electrician to deliver. Bunnings' Chief Information Officer, Genevieve Elliott, indicated that their AI assists customers with project planning, product location, stock availability, and order tracking, with continuous monitoring for helpfulness and reliability.
• Air Canada Ruling: British Columbia's Civil Resolution Tribunal ruled against Air Canada after its chatbot provided incorrect flight discount information.

  The airline's argument that the chatbot constituted a "legal entity" responsible for its own actions was rejected, and the affected traveler was compensated.

• Woolworths' Chatbot: Woolworths' chatbot, Olive, previously provided incorrect prices and engaged in irrelevant conversations, leading to adjustments in its programming. A spokesperson for Woolworths stated that customers are advised when using Olive that the system might make mistakes, and that it operates within controlled parameters with safeguards.
• UK Small Business: A small business owner in England reported that their website's AI chat offered a 25 percent discount on a large order, which the customer then negotiated up to 80 percent.

Legal Accountability and Regulatory Stance

Legal professionals and regulators consistently emphasize company responsibility for chatbot-provided information:

Matthew McMillan, who leads the digital economy practice at Lander & Rogers law firm, stated that retailers are liable for breaches under the Australian Consumer Law if a chatbot provides incorrect or misleading information.

Companies cannot attribute blame to the chatbot, as the law focuses on the impact of the conduct on consumers, irrespective of whether the message originated from a person or a machine.

The Australian Competition and Consumer Commission (ACCC) confirmed that retailers will be held accountable for information delivered by chatbots. The ACCC advised customers to report misleading information to companies or consumer protection agencies. Liabilities may arise from chatbots providing incorrect pricing, engaging in offensive content, or misrepresenting refund and return entitlements.

While some companies use disclaimers about potential chatbot errors, McMillan cautioned that these are unlikely to fully protect a company. He noted that if a chatbot provides clear, confident, but incorrect information, a background disclaimer might not mitigate the risk under law, where consumer reliance is a key factor.

The Australian federal government has shifted its approach to AI regulation, moving from a planned framework of "mandatory guardrails" under an AI act to utilizing existing laws for short-term management.

Public Trust and Expert Commentary

Experts have weighed in on the profound implications for public confidence in AI technology:

Professor Nicholas Davis from the Human Technology Institute at UTS commented that the use of AI in a deceptive manner erodes public trust.

Current AI implementations are often deployed "thoughtlessly," with an objective to provide a "nullifying response to the customer" rather than effectively solving problems.

Professor Davis advocates for establishing strict AI disclosure rules during the technology's early stages. He argues that embedding disclosure mechanisms into AI architecture now will prevent significant costs and difficulties associated with retrofitting them later.

A 2025 global study on AI trust across 17 countries positioned Australia among the most skeptical. Professor Davis clarified that this skepticism stems from public concern about how AI is being used, rather than a disbelief in its utility. He noted that Australians seek transparency and control regarding decisions made by AI systems. The Air Canada case, where an AI bot provided false information without clear identification, underscores ongoing questions about accountability in such scenarios.

Industry Practices and Varied Approaches

Retailers are implementing various strategies for deploying AI chatbots:

• Varied Responses: During recent testing, chatbots at retailers such as Myer, David Jones, The Iconic, and JB Hi-Fi either declined to answer questions about pricing and returns or offered to transfer the conversation to a human staff member.
• Specific References: The Bunnings bot was observed to engage in conversations about products while carefully referencing consumer guarantees for pricing and returns.
• Technical Issues: Kmart's bot reportedly provided error codes or generic responses.