M&A attorneys should assess communication and consent needs before a proposed AI workflow receives client information or materially influences the representation, then revisit that assessment when the use changes. The answer depends on applicable professional rules, engagement terms, client instructions, the information involved, and the tool's actual operation. ABA Formal Opinion 512 distinguishes confidentiality-related informed consent from broader communication duties under the Model Rules; it does not impose one consent requirement on every AI use. [1] Counsel should document whether the workflow requires disclosure, consultation, informed consent, modification, or rejection. Any required client discussion should be lawyer-led and completed before the affected use begins, not delegated to automated messages.
Define the proposed use before asking about consent
An assessment of 'using AI on the deal' is too broad to guide a communication decision. Describe a particular task, its inputs, the intended output, and who will rely on it. For example, an illustrative workflow might extract change-of-control language from selected contracts for an internal diligence worksheet. That is different from asking a tool to recommend a negotiating position or draft advice directly to the client. Do not treat permission for one task as permission for another.
The responsible matter attorney should own the assessment. A designated reviewer should check the output, while a qualified technology reviewer can help establish how the tool handles information. Identify the exact service, account configuration, upload method, connected systems, and proposed recipients. ABA Opinion 512 makes its confidentiality assessment depend on the client, matter, task, and tool, and keeps responsibility for professional judgment with the lawyer. [1] Establishing these facts first prevents a client conversation from resting on vague assumptions about what the software will do.
Identify governing authorities and existing client restrictions
Start with the professional rules and current guidance applicable to the representation, rather than assuming the ABA opinion governs everywhere. Opinion 512 interprets the ABA Model Rules. California's retrieved 2026 practical guidance addresses California professional obligations and expressly replaces its earlier guidance. [1] [2] Counsel should determine which authorities apply to the lawyers and matter, including any relevant proceeding-specific requirements. These sources provide an assessment framework, not a jurisdiction-wide answer for every M&A engagement. This article is general information, not advice for a particular representation.
Read the executed engagement agreement, current outside-counsel guidelines, and recorded client instructions. A client may require notice, restrict particular tools, or prohibit submitting transaction information to AI services. Under Opinion 512's Model Rule analysis, counsel must disclose AI use when the client asks or requires disclosure through engagement terms or guidelines. California's guidance similarly directs lawyers to follow applicable client instructions limiting AI use. [1] [2] A general firm approval therefore does not resolve a matter-specific restriction. If the applicable authority or an instruction is unclear, pause the proposed use and obtain qualified ethics counsel's assessment rather than asking the tool to decide.
Separate communication from confidentiality-based informed consent
Communication and informed consent are related but distinct questions. In its Model Rule 1.6 analysis, Opinion 512 concludes that entering representation information into the self-learning tools it describes requires prior informed consent because their operation creates disclosure risks. The opinion qualifies that conclusion by reference to the technology's risks and capabilities when issued. It also identifies idea generation without representation information as a use that need not require confidentiality-based consent. [1] Neither statement should become a shortcut based solely on a product label.
Separately, Opinion 512's Model Rule 1.4 analysis calls for consultation when AI output will influence a significant decision, when AI use affects the basis or reasonableness of fees, or when engagement expectations make the use material. [1] In the illustrative contract workflow, an extraction used as a checked internal aid presents a different communication question from output used to shape advice about a major transaction risk. Counsel should assess what the output actually influences. A conclusion that confidentiality consent is unnecessary does not, by itself, establish that no client communication is needed. Conversely, merely notifying a client does not establish informed consent where consent is required.
Establish the information risks before explaining them
A useful assessment identifies what information will enter the service and who could access it. Review current contractual terms, privacy policies, retention practices, training or reuse provisions, and relevant technical documentation. Investigate access by other users, other matters, provider personnel, or connected applications. Opinion 512 recommends understanding the relevant terms and obtaining expert help where needed. California's guidance says reasonable confidentiality diligence requires more than generalized marketing assurances. [1] [2] An 'enterprise' label or a claim that inputs are not used for training does not answer every storage, access, or transmission question.
For an M&A matter, identify sensitive categories such as draft transaction terms, internal legal analysis, customer contracts, and prospective buyer information. Choose the narrowest workable inputs and permissions. A limited read-only document set is a recommended starting point, not proof of confidentiality. Removing names also should not be treated as permission to submit remaining matter details without assessment. California's guidance highlights the additional risks of connected systems accessing client files and crossing matter boundaries. [2] If counsel cannot explain a material information risk, seek technical clarification or choose a different workflow. Do not promise that consent or a particular configuration preserves privilege, security, or compliance.
Make any required client discussion specific and lawyer-led
When counsel determines that disclosure or consultation is needed, explain the actual task in language the client can evaluate. Describe the proposed inputs, what the output will be used for, who remains responsible for the work, and material limitations. Explain the intended benefit without inventing time savings or promising accuracy. Discuss a realistic alternative, such as manual review, and the consequences of any client restrictions for this task. The point is to support an informed decision about the representation, not to secure a broad endorsement of AI.
Where informed consent is required under the applicable authority, the discussion needs the relevant risk particulars. Opinion 512 says the client should receive the lawyer's judgment about the purpose and benefits, the kinds of information disclosed, and how others might access or use it against the client's interests; generic engagement-letter boilerplate is insufficient. [1] Identify the authorized client decision-maker and record the scope of any decision. A staff member's informal approval or silence after an automated notice should not be treated as establishing the permission counsel needs. The attorney should conduct or supervise the discussion and review any communication before personally sending it.
Use a bounded trial with source checks and stopping conditions
Before live matter use, run a limited trial using synthetic documents or materials counsel has determined may be used. For the illustrative clause-extraction task, the accountable attorney should compare the draft worksheet with the actual contracts, amendments, and incorporated provisions. Those documents remain the source of truth. Check quoted language, document versions, references to sections, missing exceptions, and whether the tool has confused an assignment clause with a change-of-control restriction. Any legal interpretation remains the attorney's work. Opinion 512 supports task-appropriate independent verification, while California's guidance requires ongoing professional judgment and verification of AI-assisted work. [1] [2]
Set stopping conditions before the trial starts. Halt if the system accesses unapproved records, transmits information unexpectedly, invents contractual language, misses a material exception, or prevents the reviewer from tracing an output to its source. Pause live use if permissions or vendor terms change or counsel cannot complete the planned review. Record the problem and decide whether to narrow, correct, or abandon the workflow. For this framework, keep external sending, negotiation, and approvals outside the tool's permissions. A successful trial supports further assessment of this limited task; it does not establish that the tool is suitable for every document or decision.
Record the decision and define when it must be revisited
The implementation outcome is a written assessment of the communication and consent needs for the proposed workflow. Record the task, responsible attorney, governing authorities, relevant engagement provisions, client instructions, tool configuration, permitted information, material risks, and planned human checks. State the decision and reasoning: no additional communication for the narrowly defined use, disclosure or consultation, informed consent before use, modification, or rejection. This is a recommended matter-management record, not a claim that every jurisdiction prescribes this exact form.
If a client discussion occurred, record its scope, the questions addressed, and any permission or restriction in the form counsel determines appropriate under applicable rules. Keep that record with the assessment rather than treating a general firm AI policy as a substitute. Reassess when the task expands, new data categories are added, recipients change, terms change, or output begins influencing more consequential advice. California's guidance calls for evaluating communication obligations throughout the representation and reassessing evolving systems. [2] A decision for internal clause extraction should not silently extend to drafting client advice or analyzing another matter.
Keep the software boundary clear in the sale process
Client permission does not transfer legal judgment to a software provider. Counsel remains responsible for deciding the lawful scope of use and reviewing the work used in the representation. Both Opinion 512 and California's guidance emphasize retained lawyer judgment and responsibility for AI-assisted work. [1] [2] Keep client communications, transaction recommendations, and consequential decisions under accountable human control. If evidence about a tool or the applicable obligations is missing, the appropriate next step is clarification or a narrower non-AI method, not a reassuring disclosure unsupported by facts.
Acquivo is a software platform that delivers the entire sell-side M&A process, valuation, marketing materials, buyer matching, and deal management, on one platform for business owners, M&A attorneys, and boutique investment banks. It is user-operated software with no success fee, not counsel or a representative of clients. The workflow described here is a recommended assessment framework, not a statement about specific Acquivo AI features or safeguards. An attorney's documented decision should identify what the user may do, what requires further client discussion, and what remains prohibited until counsel resolves the issue.
Key Points
- Assess the exact task, tool configuration, data, and intended reliance before entering client information or using AI output in consequential advice.
- Check applicable professional authorities, engagement terms, and client instructions; ABA Model Rule guidance is not a universal jurisdictional consent rule.
- Evaluate confidentiality-based informed consent separately from disclosure and consultation obligations, and explain material risks specifically when consent is needed.
- Assign an accountable attorney, verify outputs against source records, and set trial stopping conditions before any permitted live use.
- Document the communication decision and its permitted scope, then reassess when the workflow, information, risks, or client expectations change.
Sources and references
Source retrieval dates record research access, not article publication. Read our editorial and date policy.
Choose your next step
Acquivo is user-operated software. Owners and their professionals remain responsible for decisions, communications, and professional review.
Educational information, not individual legal, tax, or investment advice. Consult qualified professionals for your circumstances. Acquivo provides software, not representation or negotiation on your behalf.