Legal Tech • October 2026

When the Client Is The Risk

Cybersecurity duties at the edge of the engagement

Written by Majo Castro

Part I of this article series argued that the biggest cybersecurity risk inside a law firm is usually framed as internal: the attorney, the staff member, or the unmanaged vendor.¹ But there is a second internal risk the ethics opinions circle without naming, because naming it is awkward: the client. Specifically, the client whose own habits undermine the confidentiality the engagement was built to protect. Sometimes that is a familiar habit, like forwarding privileged email from a personal account. Increasingly it is a newer one: the client who drops the facts of their matter into a public AI tool, handing a third party the very information the privilege depends on keeping private.

You know the pattern, even if it never shows up in breach statistics. It usually looks like one of these:

a. The client who will not touch the secure portal and insists on plain email;

b. The client who hands the engagement password to an assistant or a spouse;

c. The client who forwards privileged documents from a personal Gmail;

d. The client whose company domain has known problems that leave sensitive messages exposed; and

e. Increasingly, the client who runs the whole dispute through a free AI chatbot and forwards you the output.

None of this is exposure you chose, yet every line of it gets measured against your conduct rather than theirs. The client creates the risk, and the rule grades you.

Client Preference Is Not the Last Word
Clients are entitled to an opinion about how they communicate but not to set the security floor. American Bar Association (ABA) Formal Opinion 477R draws that line. Under Model Rule 1.6(c), you must “make reasonable efforts to prevent” inadvertent or unauthorized disclosure of client information, measured against the Comment [18] factors: how sensitive the information is, how likely disclosure becomes without safeguards, what those safeguards cost, and how much they interfere with doing the work.² The client’s preferred channel is only one input.

And once you know how a client handles privileged material, you cannot un-know it. The day you learn a client routinely forwards privileged email to a personal account, or hands out the engagement password, you are on notice. Reasonability is no longer measured from a clean slate. It is measured from what you know.

Texas landed in the same place a decade ago. Professional Ethics Committee for the State Bar of Texas Opinion 648 starts permissive: given how email works, you can generally send confidential information by email.³ Then it turns. Certain circumstances put you on the hook to warn the client and to weigh whether encryption or a different channel is smarter. Highly sensitive information going out unencrypted. An account the client shares with other people. The divorce client whose spouse knows the password.

The Duty to Warn
Confidentiality is only half the obligation. The other half is telling the client when their own setup is the problem. ABA Formal Opinion 11-459 says you “ordinarily must warn the client” when a third party could realistically reach the communications.⁴ Opinion 648 leans on that same opinion and points to Texas Disciplinary Rules of Professional Conduct (TDRPC) Rule 1.03(b) as the consultation hook.⁵ This is not a box you check once in the engagement letter. The duty re-triggers every time you learn the client’s channel is leaking the representation to someone outside it.

A warning only counts if you can prove it later. That means a documented conversation covering the security risks specific to the matter, the channels your firm uses and why, and what happens if the client routes around them. Say it, then write down that you said it.

Clients give you one of three reactions:

Acceptance. Most clients adjust. Still, document it: the conversation, confirmation of the secure channel, and matching language in the engagement letter.
Pushback. Some cite convenience or what the executive wants. Have the harder conversation before one sensitive document moves.
Regression. The toughest client agrees on the call, then slides back the moment work starts. The gap between what they promised and what they do is where the breach lives, so this is where documentation has to be tightest. It may be your only proof of what you told them.

When the Client Insists Anyway
Opinion 477R does not make you fire a client who wants ordinary email for routine work.⁶ The baseline is permissive, and the analysis builds up as stakes rise. The real question is never whether the client prefers something. It is whether this matter and this information make heightened security mandatory under the Comment [18] factors. When the answer is yes and the client still refuses, you escalate in order:

Mitigate first. Keep sensitive material off the insecure channel. Send only low-stakes pieces by email and route everything meeting the heightened-security bar through the portal.
Get informed consent. If you cannot split it cleanly, get written consent under ABA Model Rule 1.6(a) and TDRPC 1.05(c) confirming the client understands the risk and directs you to proceed. But consent has a ceiling. It cannot bless conduct that stops you from meeting the confidentiality duty at all, and it cannot cure a disclosure the client already made.
Withdraw, if it comes to that. ABA Model Rule 1.16(b) and TDRPC 1.16(b) let you withdraw when a client insists on conduct you find repugnant or fundamentally disagree with. It is the option attorneys underestimate most, and it exists for exactly this: a client who undermines the security of the representation and will not engage with mitigation or consent.

The Privilege Trap
Here is the exposure these conversations skip. A client who shares privileged communications outside the relationship can waive the privilege without you ever knowing, and a reviewing court looks hard at what the client did and whether it was reasonable.

In Semsysco GmbH v. GlobalFoundries Inc., a corporate CEO forwarded an email chain containing privileged communications to an employee of the opposing party to tee up settlement talks.⁷ His defense was that he meant to forward only the “top email.” The court found a waiver anyway. Privilege survived only if the client could show all four: that the client meant to keep the document confidential, that reasonable steps were taken to prevent disclosure, that the party moved promptly to fix it once the disclosure surfaced, and that the recipients would not be unfairly prejudiced by a protective order.

Notice where the weight sits. The whole test turns on the client’s own behavior, and “reasonable steps” include how the client handles privileged material the second it leaves your office. This is where your warnings pay off. The Opinion 11-459 conversation you documented becomes the evidence that reasonable steps existed.⁸

These exposures travel together. An ethics lapse that helped cause a waiver can resurface as Exhibit A in a malpractice claim against the lawyer who never properly advised the client. The documentation that satisfies your consultation duty is usually the same documentation that defends you later. You are not papering the file to be cautious. You are building your defense in advance.

What Can You Do Today?
Here are five questions you should be able to answer for any sensitive matter. Wherever the answer is no, that is your to-do list:

Engagement letters. Do yours say how communications get secured, which channels are required for sensitive material, and what happens if the client goes around them?
Consultation documentation. When a sensitive matter opens, is there a real conversation about secure communication and its risks, with the substance written down, not just the fact that it happened?
Channel discipline. Are attorneys and staff trained to pull communications back to secure channels when the client drifts and to document the redirect?
Informed consent. When a client overrides your security, do you have a written process capturing the risks you advised and the client’s direction to proceed anyway?
Withdrawal. Do your attorneys know when client behavior forces a withdrawal analysis under TDRPC 1.16, and is there a process for that review once mitigation and consent have failed?

The Lesson
The hardest cybersecurity conversation in your firm is not the one with the IT vendor. It is the one with the client whose habits put the engagement at risk without either of you knowing it. The rules are already on your side. You have an affirmative duty to warn under Opinion 11-459, the standing to refuse insecure practices a sensitive matter cannot absorb, and “the client insisted” has never been a defense.

MAJO CASTRO is the founder and managing attorney of Castroland Legal, a Texas-based firm concentrating on business and emerging technologies. She advises growing businesses, managed service providers (MSPs), and law firms on regulatory compliance, helping modern organizations meet rising expectations around data protection and operational accountability