How to Find AI Tools Attorneys Signed Up For With Firm Email

Why does shadow AI matter for a law firm?

Shadow AI at a law firm is any AI tool attorneys or staff use for firm work without the firm's review, and every unreviewed sign-up is a vendor that may now hold client information. A firm cannot review terms it does not know exist, so finding those tools is the first step in supervising how the firm uses AI.

California's rules put that supervision on the firm's leadership. The comment to Rule 1.1 says competence includes keeping abreast of the benefits and risks associated with relevant technology. Business and Professions Code section 6068(e)(1) requires an attorney to maintain inviolate the confidence, and preserve the secrets, of the client. Rules 5.1 and 5.3 require lawyers with managerial authority to make reasonable efforts to ensure the firm has measures giving reasonable assurance that lawyers comply with the rules and that nonlawyers' conduct is compatible with the lawyers' professional obligations.

The State Bar's practical guidance on generative AI says a lawyer must not input confidential client information into a generative AI tool that lacks adequate confidentiality and security protections, and should review the tool's Terms of Use. ABA Formal Opinion 512 adds that informed client consent is required before inputting information relating to a representation into a self-learning tool. None of that review can happen for a tool nobody has listed. This article describes what the rules and guidance say; ethics counsel should confirm how they apply to your firm.

How do attorneys sign up for AI tools with firm email?

There are three common paths, and each leaves a different trace. A "Sign in with Microsoft" or "Sign in with Google" button creates an app consent record. An email and password account using the firm address leaves a welcome or verification email. A paid plan leaves a receipt on a firm card or an expense report.

Discovery checks all three. Relying on only one leaves gaps.

How do you find AI apps connected through Microsoft or Google sign-in?

Review the list of third-party apps your users have authorized. Both Microsoft 365 and Google Workspace keep one, along with the permissions each app was granted.

• Microsoft Entra ID: under Enterprise applications, open each app's Permissions page. Microsoft shows admin consent and user consent on separate tabs, so you can see which permissions an attorney granted personally. Apps that can read mail, calendars or files deserve the closest look.
• Microsoft Defender for Cloud Apps, if your firm licenses it: the OAuth apps page rates each app's permission level as High, Medium or Low, shows how many users authorized it, and labels community use as common, uncommon or rare. A rare app with high permissions goes to the top of the review list.
• Google Workspace: in the Admin console, go to Security, Access and data control, API controls, then Manage Third-Party App Access. The Accessed apps list shows the number of users and the Google services each app requested. Google notes the list can take up to 48 hours to reflect a new grant or revocation.
• Sign-in logs: Microsoft Entra sign-in logs show which of these apps are actually in use. Entra keeps them for 7 days on Entra ID Free and 30 days on P1 or P2, and a later upgrade does not restore older logs.

How do you find AI accounts created with an email and password?

Search your mail flow for messages from AI vendors to firm addresses. Welcome, verification and billing emails show who signed up and roughly when, even when the tool never touched Microsoft or Google sign-in.

In Exchange Online, message trace searches up to 90 days of mail and accepts a wildcard sender such as *@vendor.com. Searches of 10 days or less return results almost immediately. Longer ranges produce a downloadable report that can take several hours to prepare. Running a trace requires the Organization Management role group in Exchange Online or the Exchange Administrator role in Entra.

Build a short list of vendor domains to search: the major AI chat assistants, AI legal research and drafting tools, transcription services and meeting notetakers. Then add the third trace: pull 12 months of firm card statements and expense reports and search for AI subscriptions.

What will this kind of discovery miss?

Personal accounts on personal devices. Firm records only show firm identities, firm mail and firm spending, so an attorney who uses a personal account at home will not appear.

Close that gap with two steps that do not depend on logs. First, send a short, no-blame survey asking which AI tools people use for firm work and what they use them for. Second, publish a clear list of approved tools so attorneys have a sanctioned option. Firms on Entra ID P1 or P2 can also use tenant restrictions v2 to block sign-in with consumer Microsoft accounts on firm networks and devices. AI browser extensions need their own inventory pass.

What do you do with each AI tool you find?

Approve it, replace it or block it, and record why. A dated decision log is a record of the reasonable efforts the rules describe.

• Approve: move it to a business plan with firm-managed accounts, review retention, training and confidentiality terms, and tell attorneys what information may go into it.
• Replace: if an approved tool already does the same job, move users to it and ask them to export what they need and close the old account under the vendor's terms.
• Block: revoke the app's consent in Entra ID, mark it Blocked in Google Workspace, or ban it in Defender for Cloud Apps, which can notify affected users first.

Then stop the next sign-up from being silent. Microsoft recommends allowing user consent only for apps from verified publishers and for permissions classified as low impact, with the admin consent workflow handling everything else. Google Workspace can block unconfigured third-party apps, which also stops users from signing in with Google to apps and websites the firm has not configured. Put the approved list in your firm AI policy so attorneys know where to ask.

If you learn that an attorney put client information into a tool you are blocking, treat that as a separate question for the firm's incident process and ethics counsel.

What should a law firm do about shadow AI this quarter?

Run one discovery pass, record a decision on every tool, then make the next sign-up visible. Eight steps cover it.

1. Export enterprise apps and user consents from Entra ID, and the Accessed apps list from Google Workspace if you use it.
2. Run a 90-day message trace for AI vendor domains.
3. Search 12 months of card and expense data for AI subscriptions.
4. Send a two-question survey to attorneys and staff.
5. Record an approve, replace or block decision for every tool found.
6. Restrict user consent and turn on the admin consent workflow.
7. Extend sign-in log retention beyond the default.
8. Repeat the discovery every quarter.

How s90 helps

s90's new-app alerts flag an AI tool the first time someone at the firm connects one to a firm account, and our AI guardrails help enforce the approved list. We can run the first discovery pass and hand your managing partner a list that is ready for decisions.