
These days, we’re all using AI for all sorts of things in our professional and personal lives. Most of us know that while it is a great resource and tool, we can’t necessarily rely on what it tells us to be accurate or factual.
But, when it comes to legal matters, there is a far more important reason why you shouldn’t share information or details about your matter with public AI tools like ChatGPT, Claude, Gemini, CoPilot and even Siri.
Public AI tools and chats are not confidential.
Any details or information you share with them may no longer be protected by legal privilege.
A US Federal Court recently ruled that attorney-client privilege does not apply to information or details that have been shared with a public AI tool. This is because privilege relies on confidentiality. If information is shared with a third party – whether a person or an AI tool – it is no longer confidential and therefore no longer privileged.
While we have yet to see such a case or a ruling in Australia, the non-confidential nature of public AI tools suggests that the same logic applies here.
You should never ask AI tools about anything specific to your matter, including:
- Advice your lawyer has provided you with.
- Concerns you have about the other party or yourself.
- Your legal strategy or the outcome you are hoping to achieve.
- How the law may apply to your situation.
- What a likely outcome may be.
- Communications from your lawyer.
- Drafted legal documents.
- Your personal details, or information about your children or your financial position.
- Anything you wouldn’t want the other party to know.
You can safely ask AI tools questions of a general nature, like:
- What is mediation?
- What is financial disclosure?
- What are non-financial contributions?
If you’re in doubt, don’t risk it – ask your lawyer instead.
You also shouldn’t rely on legal advice from AI tools
Any legal advice you get from an AI tool needs to be taken with a grain of salt. You see, AI tools don’t actually understand the Family Law Act and can’t interpret it or apply it to a specific situation or set of circumstances, in the way a lawyer can.
They also can’t be relied upon to distinguish between Australian law and the laws of other countries, draw on relevant precedents, assess evidence or consider other factors relevant to a case. They simply generate text that ‘sounds right’ and are known to quite literally make information up.
In fact, here in Australia, we’ve seen a handful of cases in the last few years where AI has been used to prepare legal documents and submissions that were filed with the courts and later found to cite non-existent cases and contain completely fabricated information. This has led to appeals failing, arguments being rejected, proceedings being delayed and lawyers having cost orders issued against them or being referred to the regulators for misconduct.



