Texas Lawyers Split AI Into Two Camps—and Stay Silent With Clients

Two worlds of AI in legal practice

Texas attorneys are increasingly dividing artificial intelligence tools into two distinct categories when it comes to protecting client confidentiality, according to reporting by Law.com's Texas Lawyer.

The split centers on how lawyers view AI's relationship to attorney-client privilege. Some firms are adopting what practitioners call a "walled garden" approach—using enterprise AI systems with strict data controls and confidentiality agreements. Others are taking an "open door" stance, using publicly available AI tools with fewer protections.

But the most striking finding: most Texas lawyers aren't telling clients about either approach.

The research, part three of a four-part series examining generative AI adoption among Texas attorneys, suggests the legal profession is navigating AI implementation without clear consensus on disclosure obligations. When asked whether AI threatens attorney-client privilege, lawyers' answers now depend heavily on which type of AI system they're discussing.

Why it matters

This disclosure gap creates a potential trust and ethics problem for the legal profession. As AI tools become embedded in routine legal work—from document review to research—clients may reasonably expect to know when and how their confidential information is being processed by machine learning systems. The bifurcation into "walled garden" and "open door" approaches also signals that the profession lacks unified standards for AI deployment, leaving individual firms to make consequential decisions about data security and client protection without clear regulatory guidance. For corporate legal departments evaluating outside counsel, understanding a firm's AI policies and disclosure practices may soon become a standard part of vendor assessment.

The privilege question

The core concern driving this division is whether feeding client information into AI systems compromises legal privilege. Enterprise "walled garden" solutions typically offer contractual protections, dedicated instances, and assurances that client data won't be used for model training. Public AI tools generally lack these safeguards.

Yet even as Texas lawyers make these technical distinctions internally, the research indicates most are not proactively informing clients about their AI use—regardless of which category of tools they've chosen.

This silence comes as courts and bar associations nationwide are still developing guidance on AI disclosure requirements. Some jurisdictions have begun requiring lawyers to understand the technology they use and to protect client confidentiality when deploying AI tools, but explicit disclosure mandates remain rare.

What comes next

The Texas Lawyer series provides a window into how one major legal market is adapting to generative AI in real time. The findings suggest that while lawyers are thinking carefully about security and privilege internally, client communication about these tools is lagging behind adoption.

As AI capabilities expand and become more central to legal service delivery, the gap between internal AI policies and client disclosure may become increasingly difficult to justify—particularly if clients begin asking direct questions about how their matters are being handled.

These findings were first reported by Laura Lorek for Law.com's Texas Lawyer as the third installment in a four-part investigation into AI adoption among Texas attorneys.