16 July 2026
AI chatbots are now everywhere. They answer customer questions, explain services, help with bookings, recommend products and sometimes make claims about a business.
That can be useful. However, it can also be risky.
A business can be responsible for what its AI chatbot says to customers. If a chatbot gives misleading, false or unsupported information about products, services, pricing, qualifications or legal rights, the business may not be able to avoid responsibility by saying “the AI got it wrong”.
AI is not a legal shield
Businesses are moving quickly with AI. Chatbots are being added to websites, customer service platforms, sales funnels and internal tools.
The mistake is thinking that because the answer came from a bot, the business is somehow less responsible.
That is not how risk works.
If a customer sees the chatbot as part of your business, its answers may be treated like any other business communication. That means the output needs to be accurate, clear and not misleading.
In one recent German matter, a chatbot reportedly gave incorrect information about whether health professionals were medical specialists. The business tried to explain it as a technical error. That did not get them out of trouble. The court treated the business as responsible for the chatbot’s operation and communication.
That is the point. Your chatbot is not some separate creature floating around the internet. It is part of your customer-facing business.
Courts are not blindly accepting AI answers either
There is another side to this.
In an Australian case, a court was asked to consider ChatGPT answers about the ordinary meaning of “medical centre”. The judge accepted that AI may be useful for some legal tasks, but rejected the evidence in that case.
The reasons were practical:
-
the questions lacked proper context;
-
the answers did not help with the legal construction task;
-
reliability remained a concern; and
-
it was ultimately for the court to decide the issue.
That does not mean AI is useless.
It means AI output needs to be treated carefully. It is not automatically reliable just because it sounds confident.
The real risk is ordinary customer-facing claims
Most businesses will not get into trouble because their chatbot gives a complex legal answer.
The bigger risk is simpler.
A chatbot may say:
-
a product is “Australian made” when it is not;
-
a subscription is “free” when charges may apply;
-
a service includes something that is actually excluded;
-
a professional has qualifications they do not hold;
-
a product is suitable for a particular use when that has not been checked;
-
refunds are unavailable when consumer guarantees still apply;
-
stock is available when it is not;
-
prices are fixed when extra fees apply.
Those are not “AI issues”. They are business, consumer law and brand trust issues.
The ACCC has also recently reminded businesses that consumer law applies across advertising, websites, social media, packaging, staff statements, written information and testimonials. Businesses need to ensure claims are current, correct, supported by facts and not misleading in their overall impression.
A chatbot answer is just another way a business communicates.
What businesses should do before launching an AI chatbot
Before turning on a chatbot, businesses should ask some blunt questions.
-
What is the chatbot allowed to answer?
-
What topics should it refuse or escalate?
-
Has anyone tested the answers?
-
Is it making pricing, refund, warranty or performance claims?
-
Is it giving legal, medical, financial or technical advice?
-
Does it understand the limits of the business offering?
-
Does it keep records of customer interactions?
-
Is there a human review process?
-
Who is responsible if the chatbot gives the wrong answer?
The wrong approach is to launch first and clean up later.
That may work for a rough internal experiment. It is not good enough for a public-facing customer tool.
Chatbot audits should be part of brand protection
A chatbot can damage a brand quickly.
It may misstate your services. It may promise things your team cannot deliver. It may create inconsistent customer expectations. It may expose gaps between legal terms and the way the business actually sells.
That is why chatbot review should not be left only to the tech team.
Legal, marketing, sales, customer support and management should all be involved.
At a minimum, businesses should:
-
test the chatbot using real customer questions;
-
check high-risk answers manually;
-
create a list of prohibited claims;
-
build escalation points for sensitive topics;
-
review outputs regularly;
-
update the chatbot when products, pricing or terms change;
-
keep evidence of testing and improvements.
The bottom line
AI can make business communication faster.
It can also make mistakes faster.
Businesses using AI chatbots should not assume they can blame the technology if customers are misled. If the chatbot speaks for the business, the business needs to know what it is saying.
IP Solved can help businesses review AI chatbot risk, customer-facing claims, website content and brand protection issues before they become legal problems.
