How Using AI Can Destroy Attorney Work Product Protection
Many people heard about the United States v. Heppner decision and focused on the first headline: conversations with AI tools are not protected by attorney-client privilege.
That alone should make anyone pause before sharing legal issues with AI.
But there is a second, quieter issue in the decision — one that may be even more dangerous in practice. This one affects people who already have a lawyer and believe they are simply “checking” or “researching” what counsel told them.
How Using AI Can Destroy Attorney Work Product Protection
Many people heard about the United States v. Heppner decision and focused on the first headline: conversations with AI tools are not protected by attorney-client privilege.
That alone should make anyone pause before sharing legal issues with AI.
But there is a second, quieter issue in the decision — one that may be even more dangerous in practice. This one affects people who already have a lawyer and believe they are simply “checking” or “researching” what counsel told them.
The Hidden Legal Risk in the United States v. Heppner Decision
There’s a second, not so obvious issue in the Heppner decision, one that will, we think, be much more dangerous to many more people.
In addition to ruling that conversations with AI are not protected by attorney-client privilege, the court made clear they are not protected as work product either. Work product protects materials prepared by an attorney or at an attorney’s direction in anticipation of litigation. Letters, emails, much more, produced by a lawyer or a lawyer's staff at their direction, for your case are protected.
This protection is fundamental to how legal strategy works. It allows attorneys to:
- analyze evidence
- test legal theories
- discuss possible defenses
- prepare arguments and documents
without the opposing side being able to demand those materials.
But that protection has limits.
Forwarding Your Lawyer’s Advice to AI Can Waive Work Product Protection
But - and this is a big one - if you take that email your lawyer just sent and 'run it by ChatGPT 5.2 ' just to see what it thinks, you might as well grab the mic on opening day for the Mariners and broadcast it to all 48,000 at T-Mobile Park.
This is simple, Judge Rakoff ruled that relaying what your lawyer told you to a public AI tool eviscerates work product protection.
In other words, once you voluntarily share that information with a third party — including an AI system — the protection surrounding that legal analysis can disappear.
Why People Share Their Lawyer’s Advice With AI Without Realizing the Risk
It’s the most natural thing in the world: you paraphrase counsel, you ask the bot to “explain it,” you copy an email string, you ask whether your lawyer is right 'on this one point' or why they don't care about 'evidence you have', or a hundred other things while you look for more affirmation during what has to be the most stressful time of your life.
You'll probably get that affirmation - that's what AI is really good at unless you tell it different - but at a huge cost: you've just accidentally built a discoverable record out of what were privileged conversations.
People paste into AI the very thing they most need protected—what their lawyer told them. Once that advice is shared with a third party — and an AI chatbot is a third party — it is no longer confidential.
Sharing Legal Strategy With Third Parties Can Destroy Legal Protections
While we're at it - the same principle applies when you tell your brother-in-law “who knows a lot about this stuff.” Or your friend who had a 'legal problem' a few years back and really worked on his own defense. Or anyone else outside the attorney-client relationship. Work product does not survive sharing. Online or off.
This is not a technical loophole. It is how the legal system has always worked.
Once confidential legal material leaves the protected attorney-client relationship, courts may treat it as voluntarily disclosed.
Why AI Cannot Understand the Full Context of Your Legal Case
One last thing: AI does not and never will know all the facts of your case. know your file. It does not know what was said off the record. It does not know the judge. It does not know the prosecutor. It does not know what has already been disclosed and what has not. It generates answers based on patterns, not context.
So now you have two problems: you may have waived protection, and you have injected analysis into your case that is built on incomplete information.
This combination — lost protection and flawed analysis — is exactly the kind of situation prosecutors and opposing counsel look for during litigation.
If You Have Questions About Your Case, Ask Your Lawyer — Not AI
None of this means you should not ask questions. You should ask them — to your lawyer. That is what privilege exists for. That is what work product exists for. That is what strategy sessions are for.
Legal strategy is not something to crowdsource to software, search engines, or well-meaning relatives.
It is something to discuss confidentially with the lawyer responsible for protecting you.
The Right to Counsel — Not AI Counsel
AI can summarize information. It can explain legal terminology. It can point to general concepts.
What it cannot do is protect your rights, maintain privilege, or safeguard your legal strategy.
Only a lawyer can do that.
At Knauss Law, we believe every person facing criminal allegations deserves something far more reliable than a chatbot:
the right to experienced, confidential legal counsel.
Not AI counsel.
Real counsel.
If you are facing criminal charges, an investigation, or even early questions about your case, talk to your lawyer before sharing information with anyone — including an AI system.
Because once strategy leaves the attorney-client relationship, getting it back can be impossible.
425-300-5566