---
title: "Can a law firm share AI memory across matters?"
description: "Firm know-how without client information, yes. Client information across matters needs a risk assessment, separation per matter and, often, client consent."
image: "https://bombos.ai/assets/og-en-v2.png"
---

# Can a law firm share AI memory across matters and lawyers?

Short answer

Firm knowledge without client information, like workflows, templates and researched legal questions, can be shared. Client information belongs to the matter. In the US, ABA Formal Opinion 512 (July 29, 2024) says that for self-learning AI tools that can disclose it, informed client consent is needed first, even if only lawyers at the same firm use the tool.

*Piet Baudoin · published November 16, 2026 · November 16, 2026*

## Two kinds of knowledge, two answers

A law firm holds knowledge everyone should have and knowledge that belongs to one client. For AI memory, that is the whole difference.

| Kind of knowledge | Examples | Share across matters and lawyers? |
| --- | --- | --- |
| Firm know-how without client information | Workflows, templates, style, legal research in general form | Yes |
| Information relating to the representation | Facts, documents, strategy, what the client said | Stays with the matter; use beyond it needs informed consent and your own analysis |
| Information from former or prospective clients | A closed matter, an intake that went nowhere | Also confidential; the opinion points to Model Rules 1.9(c) and 1.18(b) |

## What ABA Formal Opinion 512 says

ABA Formal Opinion 512, issued July 29, 2024, is advisory, not binding, and written for US lawyers. Its reasoning still maps onto any firm considering shared AI memory:

1.  Confidentiality covers “all information relating to the representation of a client, regardless of its source”.
2.  Before use, the lawyer has to assess the risk that people outside the firm could get access to the information.
3.  With self-learning tools, “a client’s informed consent is required prior to inputting information relating to the representation”.
4.  The risk exists “even if the tool is used exclusively by lawyers at the same firm.”
5.  General boilerplate in an engagement letter is not informed consent.
6.  Without client information, for example when working through ideas, no consent is needed.

Point 4 is the core for this question. The opinion is about tools that learn from what goes in and can surface it later. A memory that stores facts from matter A and can retrieve them in matter B raises the same risk, but whether it does depends on how it is built: who can retrieve what, and whether it is separated per matter. That is the assessment the opinion asks for.

Outside the US, the frame differs per bar. The Netherlands Bar (NOvA) published a revised version of its “Aanbevelingen AI in de advocatuur” in November 2026, and it addresses this question directly. Recommendation E.6 asks the lawyer to assess whether an AI system with access to information from several clients, for example through shared context windows, memory functions or fine-tuning on firm-wide data, creates a risk of information flowing between clients, and to put technical or organizational safeguards in place where it does. It also asks firms to check how separation between client files is secured when buying such a system, and to include it in the regular conflict-of-interest check. Recommendation D.1 asks firms to inform clients about their AI policy by default and to ask consent in specific cases, for example when professional secrecy cannot otherwise be guaranteed.

## A decision route for each piece of knowledge

1.  **Does it contain information relating to a client?** If not (a template, a research memo in general form), it can go into firm-wide memory.
2.  **If it does, which matter does it belong to?** Store it only in that matter’s memory, readable only by the people on that matter.
3.  **Could two clients of the firm end up adverse?** Then the conflict check also covers what the AI memory can reach.
4.  **Does the tool meet the firm’s terms?** Where data is stored, whether it trains models, and who at the vendor can see it.
5.  **Is consent needed?** Under your own bar’s rules; under the NOvA recommendations, at least when secrecy cannot otherwise be guaranteed.

## The misconception: “it stays inside the firm, so it is safe”

That addresses whether an outsider can read along. It does not address whether information from one matter may surface in another. A colleague on a related matter for a different client, or a later matter adverse to a former client, should not receive that information through an AI answer. Memory that works per person does not make that distinction.

In the common tools, memory is also tied to the person. In Claude, each project has its own memory, but a user’s memory from chats outside projects also flows into every project unless Use account memory is turned off for that project. Owners can’t view or edit a user’s individual memories in the app; memory does appear in data exports. A lawyer who works on matters outside projects carries one memory across them, and the firm can’t manage it day to day.

## What this means for the firm

Shared memory can do a lot for firm know-how: how you structure a brief, which legal question is already researched, which mistakes not to repeat. For client information the design requirement is the reverse: separated per matter, access per matter, and a visible source for each answer so you can check it did not come from another matter. For how an agency sets up the same separation per client in Claude, see [shared AI memory per client for a marketing agency](https://bombos.ai/en/questions/by-team/marketing-agency-shared-ai-memory-per-client/). On capturing firm knowledge for AI in general: [how to give AI agents your company context](https://bombos.ai/en/knowledge-base/how-to-give-ai-agents-your-company-context/).

## Where Bombos fits

Bombos is a plugin that gives every AI tool in a firm one shared memory, with knowledge and learned rules that can apply only within their own context, such as a single matter, and with each answer showing what it rests on. It can run on your own server with your own models, which is often the first question with confidential material. Whether and how client information is used remains the firm’s call; one lawyer can try it first for their own agents.

## Related questions

Is ABA Formal Opinion 512 binding?

No. ABA formal opinions are advisory; what binds a lawyer are the rules of their own jurisdiction. Opinion 512 is still a clear statement of where the risks are.

Is a clause in the engagement letter enough consent?

Not according to Opinion 512. It says general boilerplate language in an engagement letter does not amount to informed consent; the client needs to understand what goes into the tool and what the risks are.

Do I need consent if I enter no client information?

According to the opinion, no. Using a tool for general ideas or questions that contain no information relating to a representation does not require consent.

Can Claude keep memory separate per matter?

Partly. Each Claude project has its own memory, and you can turn off Use account memory per project. Sharing is per person with Can view or Can edit, on Team or Enterprise.

## Sources

1.  [The Bar Examiner: Generative Artificial Intelligence Tools (on ABA Formal Opinion 512)](https://thebarexaminer.ncbex.org/article/fall-2024/generative-artificial-intelligence-tools/)
2.  [ABA Formal Opinion 512: Generative Artificial Intelligence Tools](https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf)
3.  [NOvA: Aanbevelingen AI in de advocatuur, herziene versie november 2026](https://www.advocatenorde.nl/sites/default/files/2026-11/Aanbevelingen%20AI%20in%20de%20advocatuur%202026%20DEF.pdf)
4.  [Claude Help Center: How can I create and manage projects?](https://support.claude.com/en/articles/9519177-how-can-i-create-and-manage-projects)
5.  [Claude Help Center: Use Claude's chat search and memory to build on previous context](https://support.claude.com/en/articles/11817273-use-claude-s-chat-search-and-memory-to-build-on-previous-context)

## What does your AI forget?

1.  01You tell us which AI you use and where it keeps starting over.
2.  02We show what a shared memory changes there, on your own examples.
3.  03You keep the plan, with or without Bombos.

![Piet Baudoin](https://bombos.ai/assets/team/piet-klein.jpg)

**Piet Baudoin**You talk to Piet, from the Bombos teamDutch team from Utrecht and Nijmegen · data in Europe

## More questions about By type of team

-   [How does an accounting firm share AI memory without mixing up client files?](https://bombos.ai/en/questions/by-team/accounting-firm-ai-memory-per-client/)
-   [Can a marketing agency share Claude Projects and AI memory across its team, client by client?](https://bombos.ai/en/questions/by-team/marketing-agency-shared-ai-memory-per-client/)
-   [Can a healthcare team share AI projects and memory when patient data is involved?](https://bombos.ai/en/questions/by-team/healthcare-team-shared-ai-memory-patient-data/)
-   [How do you share the same AI context across a dev team using Claude Code, Cursor and Copilot?](https://bombos.ai/en/questions/by-team/share-ai-context-dev-team-claude-code-cursor-copilot/)
-   [How can a consulting firm reuse what it learned on past engagements with AI, without mixing client data?](https://bombos.ai/en/questions/by-team/consulting-firm-reuse-ai-knowledge-across-clients/)
-   [How does a sales team give every rep's AI the same account context?](https://bombos.ai/en/questions/by-team/sales-team-shared-ai-account-context/)
-   [How does a support team get its AI to use what agents already solved in past tickets?](https://bombos.ai/en/questions/by-team/support-team-ai-knowledge-from-past-tickets/)

[All questions about By type of team →](https://bombos.ai/en/questions/by-team/)


---

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://bombos.ai/#organization","name":"Bombos","url":"https://bombos.ai/","logo":"https://bombos.ai/assets/bombos-bee.png","description":"Bombos is the memory layer under every AI your team uses. One shared memory for Claude, ChatGPT, Copilot, Cursor and your own agents, that learns how you work.","sameAs":["https://www.linkedin.com/company/bombos-ai","https://www.youtube.com/@truebombos","https://www.instagram.com/truebombos/","https://bsky.app/profile/bombostrue.bsky.social","https://www.threads.com/@truebombos","https://mastodon.social/@Bombos","https://www.tiktok.com/@truebombos","https://t.me/truebombos","https://www.pinterest.com/truebombos/","https://github.com/bombos-ai","https://linktr.ee/bombos.ai","https://bombosai.substack.com","https://hashnode.com/@bombosai","https://hackernoon.com/u/bombos","https://dev.to/bombos"],"email":"info@bombos.ai","areaServed":["NL","Europe"],"knowsLanguage":["nl","en"],"contactPoint":{"@type":"ContactPoint","email":"info@bombos.ai","contactType":"sales","url":"https://bombos.ai/en/contact/","availableLanguage":["nl","en"]}},{"@type":"WebSite","@id":"https://bombos.ai/#website","name":"Bombos","url":"https://bombos.ai/","inLanguage":["nl-NL","en"],"publisher":{"@id":"https://bombos.ai/#organization"}}]}

{"@context":"https://schema.org","@type":"Article","headline":"Can a law firm share AI memory across matters and lawyers?","description":"Firm know-how without client information, yes. Client information across matters needs a risk assessment, separation per matter and, often, client consent.","abstract":"Firm knowledge without client information, like workflows, templates and researched legal questions, can be shared. Client information belongs to the matter. In the US, ABA Formal Opinion 512 (July 29, 2024) says that for self-learning AI tools that can disclose it, informed client consent is needed first, even if only lawyers at the same firm use the tool.","datePublished":"2026-11-16T00:00:00.000Z","dateModified":"2026-11-16T00:00:00.000Z","author":{"@type":"Person","name":"Piet Baudoin"},"publisher":{"@type":"Organization","@id":"https://bombos.ai/#organization","name":"Bombos","logo":{"@type":"ImageObject","url":"https://bombos.ai/assets/bombos-bee.png"}},"image":"https://bombos.ai/assets/og-en-v2.png","mainEntityOfPage":"https://bombos.ai/en/questions/by-team/can-law-firm-share-ai-memory-across-matters/","inLanguage":"en-US","isPartOf":{"@type":"CollectionPage","name":"By type of team","url":"https://bombos.ai/en/questions/by-team/"},"citation":[{"@type":"CreativeWork","name":"The Bar Examiner: Generative Artificial Intelligence Tools (on ABA Formal Opinion 512)","url":"https://thebarexaminer.ncbex.org/article/fall-2024/generative-artificial-intelligence-tools/"},{"@type":"CreativeWork","name":"ABA Formal Opinion 512: Generative Artificial Intelligence Tools","url":"https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf"},{"@type":"CreativeWork","name":"NOvA: Aanbevelingen AI in de advocatuur, herziene versie november 2026","url":"https://www.advocatenorde.nl/sites/default/files/2026-11/Aanbevelingen%20AI%20in%20de%20advocatuur%202026%20DEF.pdf"},{"@type":"CreativeWork","name":"Claude Help Center: How can I create and manage projects?","url":"https://support.claude.com/en/articles/9519177-how-can-i-create-and-manage-projects"},{"@type":"CreativeWork","name":"Claude Help Center: Use Claude's chat search and memory to build on previous context","url":"https://support.claude.com/en/articles/11817273-use-claude-s-chat-search-and-memory-to-build-on-previous-context"}]}

{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Bombos","item":"https://bombos.ai/en/"},{"@type":"ListItem","position":2,"name":"Questions","item":"https://bombos.ai/en/questions/"},{"@type":"ListItem","position":3,"name":"By type of team","item":"https://bombos.ai/en/questions/by-team/"},{"@type":"ListItem","position":4,"name":"Can a law firm share AI memory across matters and lawyers?","item":"https://bombos.ai/en/questions/by-team/can-law-firm-share-ai-memory-across-matters/"}]}

{"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"Is ABA Formal Opinion 512 binding?","acceptedAnswer":{"@type":"Answer","text":"No. ABA formal opinions are advisory; what binds a lawyer are the rules of their own jurisdiction. Opinion 512 is still a clear statement of where the risks are."}},{"@type":"Question","name":"Is a clause in the engagement letter enough consent?","acceptedAnswer":{"@type":"Answer","text":"Not according to Opinion 512. It says general boilerplate language in an engagement letter does not amount to informed consent; the client needs to understand what goes into the tool and what the risks are."}},{"@type":"Question","name":"Do I need consent if I enter no client information?","acceptedAnswer":{"@type":"Answer","text":"According to the opinion, no. Using a tool for general ideas or questions that contain no information relating to a representation does not require consent."}},{"@type":"Question","name":"Can Claude keep memory separate per matter?","acceptedAnswer":{"@type":"Answer","text":"Partly. Each Claude project has its own memory, and you can turn off Use account memory per project. Sharing is per person with Can view or Can edit, on Team or Enterprise."}}]}
```
