Ethical Walls in the Age of AI

Clients tell their lawyers things they wouldn’t tell their spouse. Pending acquisitions. Litigation strategy. Regulatory exposure. All of it shared with the expectation that it stays locked to the matter team. 

Ethical walls enforce that expectation. They keep conflicts of interest from turning into breaches of duty. 

Here’s what’s uncomfortable… most of the AI tools entering legal workflows right now have no idea those walls exist. 

The gap between AI hype and firm reality 

General-purpose AI assistants, including the copilots bundled with Microsoft 365, index broadly. They surface documents, emails, contacts, and conversations based on relevance and recency. 

That works fine until your firm has an active ethical wall between two matters. 

A generic copilot doesn’t know that Attorney A is walled off from Matter B. It doesn’t know that the research memo in the DMS is off-limits to half the corporate group. It doesn’t check conflict data before returning a search result. 

This is not a theoretical risk. If you’re adopting AI at your firm and your tools can’t see your walls, you have a compliance problem whether you’ve noticed it yet or not. 

Why bolt-on wall enforcement breaks 

Some firms have tried layering ethical wall logic on top of general-purpose tools after deployment. It doesn’t hold. 

The issue is structural. Unless the platform consumes wall data at the API level, you end up with a patchwork of manual exclusions and user-level permissions. That patchwork breaks the moment someone changes roles, a matter shifts, or a new wall goes up. 

Wall enforcement needs to live in the platform, not on top of it. It has to run through search, directory lookups, document rendering, and AI responses. If it’s added after the fact, it leaks. 

How Infodash approaches this 

We built Infodash for law firms, so ethical walls were a design constraint from the start. Not something we patched in later. 

In practice, it works across several layers: 

Search checks wall data from Intapp or iManage SPM before returning results. Walled content doesn’t show up with a “restricted” label. It just doesn’t show up. The user sees a clean result set and has no indication anything was filtered. 

Directories respect wall boundaries too. If a user is walled off from a matter, the associated personnel connections are filtered out of listings. 

The API enforces walls at the data layer, not the UI. That’s a meaningful distinction. It means every integration built on Infodash, whether it’s a Power Automate workflow, a Teams app, or a custom dashboard, inherits wall compliance automatically. You don’t rebuild wall logic for each new surface. 

DashBots (to be released Q3 2026), our AI assistants on Azure, work through permissioned function calls. The LLM never touches firm data directly. Every query routes through the Infodash API, so every response respects the same wall boundaries as a normal search. The AI sees what the user is allowed to see. Nothing more. 

And everything is logged. Searches, API calls, DashBot interactions, what was filtered and why. Compliance teams can pull a record for any query and see that walls were enforced. 

What to ask your vendors 

If your firm is evaluating AI for knowledge management or enterprise search, ethical wall enforcement should be high on the list. Not buried in a compliance appendix. 

Five questions worth asking: 

1. Does your platform consume ethical wall data natively, or does it rely on manual configuration? 

2. Is wall enforcement applied at the API level, or only at the UI? 

3. When a new wall goes up, how quickly does the system reflect it? 

4. Can you show me an audit trail proving wall enforcement on a specific query? 

5. If I build a custom integration on your API, does it inherit wall enforcement? 

The answers tell you a lot about whether you’re looking at a platform built for law firms or a general-purpose tool with a legal label on it. 

Have a question? Need a quote? Looking for a demo?
Click here to start a conversation with us.