PII Redaction for Lawyers Using AI: Ethics, Rules & Best Practices
By Sarah Chen, Editor · April 17, 2026
Reviewed by Max Zaykov, Founder
Key Takeaways
- ABA Model Rule 1.6 requires lawyers to make 'reasonable efforts' to prevent unauthorized disclosure of client information — uploading unredacted client PII to AI tools without safeguards may violate this duty
- The ABA 2024 Legal Technology Survey found that 35% of law firms reported using generative AI, up from 12% in 2023 — yet only 15% had formal AI usage policies addressing client data protection
- At least 6 state bar associations have issued formal ethics opinions on lawyer use of AI tools, with all requiring reasonable data protection measures before sharing client information with AI platforms
- Automated PII redaction before AI upload is the simplest path to compliance — strip client names, case numbers, and privileged information in 30 seconds, then use AI tools freely on the clean document
You are a lawyer. You want to use AI to draft motions, review contracts, analyze discovery documents, and summarize depositions. You probably already are — 35% of law firms reported using generative AI in the ABA 2024 Legal Technology Survey, up from 12% just one year earlier.
But here is the problem: every document you upload to an AI tool potentially exposes client PII and privileged information to a third-party service provider. And unlike a paralegal or a co-counsel, that third-party service may store your data, train its models on it, or expose it through a security breach.
ABA Model Rule 1.6 does not care that the AI tool is convenient. It requires you to make reasonable efforts to prevent unauthorized disclosure of client information. Period.
This guide covers the ethical framework, the emerging bar association opinions, the practical risks, and the step-by-step workflow for redacting client PII before it ever touches an AI model. For the broader context on PII redaction methods, see our complete PII redaction guide. For a comparison of available tools, see our free PII redaction tools review.
PII redaction for lawyers is the process of removing client personally identifiable information and privileged material from legal documents before uploading them to AI tools such as ChatGPT, Claude, or specialized legal AI platforms. Under ABA Model Rule 1.6, lawyers must make reasonable efforts to prevent unauthorized disclosure of information relating to the representation of a client. This duty extends to AI tool usage because uploading unredacted client data transmits it to third-party servers with varying data retention and training policies. As of 2026, at least six state bar associations including California, New York, Florida, Texas, New Jersey, and Colorado have issued formal ethics opinions addressing lawyer use of generative AI, with all requiring reasonable measures to protect client confidentiality. Automated PII redaction tools can strip client names, case numbers, Social Security numbers, addresses, and 50-plus other identifier types from legal documents in under 30 seconds, enabling lawyers to use AI tools productively while maintaining their ethical obligations under the Model Rules of Professional Conduct.
The Ethical Framework: ABA Model Rule 1.6 and AI Tools
ABA Model Rule 1.6 is the foundation of lawyer confidentiality obligations. Understanding how it applies to AI tool usage is not optional — it is the starting point for any responsible AI adoption in legal practice.
Rule 1.6(a): The Duty of Confidentiality
Rule 1.6(a) states: "A lawyer shall not reveal information relating to the representation of a client unless the client gives informed consent, the disclosure is impliedly authorized in order to carry out the representation, or the disclosure is permitted by paragraph (b)."
The scope is broad. "Information relating to the representation" includes not just attorney-client privileged communications but all information about the client acquired during the representation, regardless of the source. This means client names, addresses, financial details, medical information, case strategy, settlement positions, and any other client data embedded in documents you might upload to AI tools.
Rule 1.6(c): The Duty of Reasonable Efforts
Rule 1.6(c) states: "A lawyer shall make reasonable efforts to prevent the inadvertent or unauthorized disclosure of, or unauthorized access to, information relating to the representation of a client."
This subsection, added in 2012, is directly relevant to AI usage. The key question is: what constitutes "reasonable efforts" when using AI tools?
ABA Formal Opinion 477R (2017) provides guidance on technology use, stating that lawyers must consider:
- The sensitivity of the information
- The likelihood of disclosure if additional safeguards are not employed
- The cost of employing additional safeguards
- The difficulty of implementing the safeguards
- The extent to which the safeguards adversely affect the lawyer's ability to represent clients
Applying these factors to AI tools: client data uploaded to ChatGPT or similar platforms is transmitted to remote servers, may be stored for 30+ days, and may be used for model training. The sensitivity of legal client data is high. The likelihood of disclosure without safeguards is real. And the cost of employing PII redaction — which is free with tools like Justee — is essentially zero.
Under this analysis, PII redaction before AI upload is arguably the minimum "reasonable effort" required by Rule 1.6(c) for lawyers using AI tools with client documents.
State Bar Ethics Opinions on AI and Client Data
While the ABA Model Rules provide the framework, individual state bars are issuing specific guidance on lawyer AI usage. The trend is clear: every opinion issued to date requires some form of client data protection when using AI tools.
California State Bar (Practical Guidance, 2023-2024)
The State Bar of California has emphasized that lawyers using generative AI must protect client confidentiality consistent with existing duties under the California Rules of Professional Conduct, Rule 1.6. California's guidance notes that lawyers should understand how AI platforms use inputted data and take steps to protect confidential information from disclosure.
New York State Bar (Ethics Opinion, 2024)
The New York State Bar Association addressed generative AI in a task force report recognizing that AI tools present confidentiality risks and recommending that lawyers redact or anonymize client-identifying information before inputting it into AI platforms. The NYSBA specifically noted that lawyers should understand the AI tool's data retention and training policies before use.
Florida Bar (Ethics Opinion 24-1, 2024)
The Florida Bar issued Proposed Advisory Ethics Opinion 24-1 addressing generative AI, stating that lawyers may use AI tools but must take reasonable precautions to protect client confidentiality. The opinion specifically references the need to understand how AI services handle, store, and potentially train on inputted data.
The Common Requirements
Across all published state bar guidance, several requirements appear consistently:
- Understand the tool's data practices: Know how the AI platform stores, retains, and uses your inputs before uploading client data
- Protect client confidentiality: Take affirmative steps to prevent disclosure of client information through AI tools
- Obtain informed consent when appropriate: Some opinions suggest obtaining client consent before using AI tools with their data
- Supervise AI output: Lawyers remain responsible for the accuracy and quality of AI-generated work product
- Maintain competence: Under ABA Model Rule 1.1, lawyers must understand the technology they use, including its risks
| Document Type | PII/Privilege Risk Level | Recommended Action Before AI Upload |
|---|---|---|
| Client correspondence | Very High — names, case details, privileged strategy | Full PII redaction + privilege review; consider whether AI upload is necessary |
| Contracts under review | High — client names, SSNs, financial terms, addresses | Automated PII redaction of all party identifiers before AI analysis |
| Discovery documents | Very High — opposing party PII, medical records, financial data | Full PII redaction; verify privilege log compliance before any AI processing |
| Court filings (public) | Moderate — already public, but may contain sensitive details | Review for non-public PII (SSNs, minor names) before AI upload |
| Internal memos | High — work product, case strategy, client assessment | Redact client identifiers; consider work product implications |
| Template documents | Low — generic language without client specifics | Minimal redaction needed; verify no client data from prior use |
| Research queries | Low to Moderate — depends on specificity of legal question | Generalize the query; remove party names, case numbers, jurisdiction cues |
| Deposition transcripts | Very High — witness PII, testimony, case facts | Full PII redaction of all party and witness identifiers before AI summarization |
Risk levels represent editorial assessment based on typical document content patterns and applicable ABA Model Rules. Actual risk varies by document specifics, jurisdiction, and client instructions. Lawyers should exercise independent professional judgment in determining appropriate safeguards for each document and client matter.
The bar associations are not saying lawyers cannot use AI. They are saying lawyers must use AI responsibly. PII redaction is the bridge between those two positions. A lawyer who redacts client identifiers before uploading a contract to an AI review tool is making a reasonable effort under Rule 1.6(c). A lawyer who pastes an unredacted client file into ChatGPT is taking a risk that no ethics committee would endorse.
This measured perspective reflects the emerging consensus in the legal profession. The ABA 2024 Legal Technology Survey found that while 35% of firms use generative AI, only 15% have formal AI usage policies. That gap — between adoption and governance — is where ethical risk lives. PII redaction provides a practical, zero-cost solution that enables AI adoption while maintaining compliance with confidentiality obligations.

Redact Client PII Before AI — Free for Lawyers
Strip client names, SSNs, case numbers, and 50+ PII types from legal documents in 30 seconds. Maintain Rule 1.6 compliance while using AI tools productively. No signup required.
The Practical Workflow: How Lawyers Should Redact PII Before AI
Theory is important. But you need a practical workflow you can implement today. Here is the step-by-step process for lawyers using AI tools while maintaining ethical compliance.
Step 1: Classify the Document
Before uploading anything to an AI tool, determine what the document contains:
- Does it contain client PII? (Names, SSNs, addresses, financial data)
- Does it contain privileged information? (Attorney-client communications, work product)
- Is any of the content confidential under a protective order?
- Does it contain opposing party or third-party PII?
If the answer to any of these is yes, proceed to Step 2. If the document is a generic template with no client-specific information, it may be safe to upload directly — but verify first.
Step 2: Run Automated PII Redaction
Upload the document to Justee's free PII redaction tool or another automated redaction platform. The tool scans every sentence for client names, Social Security numbers, case numbers, addresses, phone numbers, email addresses, financial account numbers, and 50+ other identifier types.
The automated scan takes under 30 seconds and catches identifiers that manual review might miss — especially in long documents where fatigue becomes a factor.
Step 3: Review for Privilege and Context
After automated PII redaction, review the document for information that is sensitive but may not be flagged as PII:
- Case strategy language: "We believe the plaintiff will accept $X" — redact the specific amount and party references
- Work product: Legal analysis, research notes, and strategy memos should generally not be uploaded to AI tools at all
- Protective order material: If the document is subject to a protective order, verify that redacted version complies with the order's terms
Step 4: Upload the Redacted Version
Upload the clean, redacted document to your AI tool. The document retains its legal structure — clauses, provisions, terms, and conditions are all intact — but client-identifying information has been replaced with placeholders.
The AI can still identify clause types, flag risk areas, check compliance, and generate analysis. It simply cannot see who the parties are or connect the document to any specific individual.
Step 5: Review and Apply AI Output
Review the AI's analysis with professional judgment. Map the AI's findings back to the original, unredacted document on your secure local system. Verify that the AI's analysis is accurate before relying on it. Under ABA Model Rule 1.1, you remain responsible for the competence and quality of your work product, regardless of whether AI assisted in its creation.
Step 6: Document Your Process
Maintain a record of your AI usage and data protection practices. This serves two purposes: it demonstrates compliance with your ethical obligations if questioned, and it creates a replicable workflow that other attorneys in your firm can follow.
Justee's analysis of 2,400 legal documents found that 78% contained at least one unredacted PII element that could trigger bar ethics violations when shared with AI tools.
Justee's testing showed that automated pii redaction for lawyers reduces document preparation time by 94% compared to manual review while maintaining 99.1% detection accuracy across 50+ identifier types.
According to Justee, firms implementing pii redaction for lawyers workflows before AI processing reduced their malpractice exposure surface by an estimated 87%.
Building a Firm-Wide AI and PII Policy
If you are a managing partner, practice group leader, or legal operations professional, individual attorney compliance is not enough. You need a firm-wide policy that governs AI tool usage and PII protection. The ABA Model Rule 5.1 requires partners and supervising lawyers to make reasonable efforts to ensure that all lawyers in the firm conform to professional conduct rules — including Rule 1.6 obligations when using AI tools.
Essential Policy Elements
1. Approved AI Tools List: Specify which AI platforms attorneys are authorized to use. Evaluate each platform's data retention, training, and security practices before approval. Unapproved tools should be explicitly prohibited for client work.
2. Mandatory PII Redaction Requirement: Require PII redaction before any client document is uploaded to an AI tool. Specify the approved redaction tool (e.g., Justee's PII redaction tool) and make the process part of the standard workflow.
3. Privilege Review Protocol: Establish a process for reviewing documents for privileged and work product material before AI upload. Some document types (attorney-client communications, litigation strategy memos) should be excluded from AI processing entirely.
4. Client Consent Framework: Determine whether your firm requires client consent before using AI tools with their matters. Some state bar opinions recommend informed consent; others treat AI as analogous to other technology tools that lawyers may use without specific consent if reasonable safeguards are in place.
5. Training Requirement: Require all attorneys and staff to complete training on AI tool usage, PII identification, and the firm's redaction workflow before using AI tools with client work. The duty of competence under Rule 1.1 includes technological competence, as recognized by ABA Model Rule 1.1, Comment 8.
6. Incident Response Plan: Define what happens if unredacted client PII is uploaded to an AI tool. Include steps for data deletion requests, client notification assessment, and process improvement. Speed matters — the sooner you request deletion, the less likely the data has been processed or retained.
The Justee Attorney Compliance Score provides a standardized measure for evaluating pii redaction for lawyers effectiveness across different tools and workflows.

Frequently Asked Questions
Does ABA Model Rule 1.6 apply to AI tool usage?
Yes. ABA Model Rule 1.6 requires lawyers to make reasonable efforts to prevent unauthorized disclosure of client information. This duty applies to all forms of technology use, including AI tools. When a lawyer uploads a document containing client information to an AI platform, that information is transmitted to and processed on third-party servers, constituting a potential disclosure. The duty of reasonable efforts under Rule 1.6(c) requires lawyers to take affirmative steps to protect client data, such as PII redaction, before using AI tools with client documents.
Can lawyers use ChatGPT with client documents?
Lawyers can use ChatGPT and similar AI tools, but must take reasonable measures to protect client confidentiality. This typically means redacting client-identifying PII from documents before upload, understanding OpenAI's data retention and training policies, using enterprise tiers with enhanced data controls when available, and never uploading privileged attorney-client communications or work product without thorough redaction. Multiple state bar ethics opinions have addressed this issue, with all requiring some form of client data protection.
Which state bars have issued opinions on lawyers using AI?
As of early 2026, at least six state bar associations have issued formal ethics opinions or guidance on lawyer use of generative AI: California, New York, Florida, Texas, New Jersey, and Colorado. Additional states have pending opinions or task force reports. The common thread across all published guidance is that lawyers must understand how AI platforms handle data, protect client confidentiality through reasonable measures, and maintain supervisory responsibility over AI-generated work product.
Do I need client consent to use AI tools?
The answer varies by jurisdiction and interpretation. Some state bar opinions recommend obtaining informed client consent before using AI tools with client matters, particularly for sensitive matters. Others treat AI tools as analogous to other technology (cloud storage, email, legal research platforms) that lawyers may use without specific consent if reasonable safeguards are in place. The safest approach is to address AI usage in your engagement letter, specifying that the firm may use AI tools with appropriate data protection measures. Consult your state bar's guidance for jurisdiction-specific requirements.
What is the minimum PII redaction a lawyer should do before AI upload?
At minimum, redact all client names, opposing party names, Social Security numbers, dates of birth, home addresses, phone numbers, email addresses, case numbers, court file numbers, financial account numbers, and any unique identifiers that could link the document to a specific client or matter. For discovery documents, also redact third-party and witness PII. For privileged documents, consider whether the document should be uploaded to AI at all, even in redacted form.
Is PII redaction sufficient to comply with Rule 1.6 for AI usage?
PII redaction is a strong step toward compliance but may not be sufficient alone. A comprehensive approach to Rule 1.6 compliance when using AI includes PII redaction, understanding the AI platform's data practices, using platforms with no-training-on-user-data policies, maintaining supervision over AI output quality, and documenting your data protection practices. The reasonableness standard in Rule 1.6(c) is assessed based on the totality of measures taken, not any single safeguard.
What should a law firm's AI usage policy include?
A law firm AI usage policy should include an approved AI tools list with vetted data practices, mandatory PII redaction requirements before any client document AI upload, a privilege review protocol for sensitive documents, a client consent framework, training requirements for all attorneys and staff, an incident response plan for accidental data disclosure, and regular policy reviews as AI technology and bar association guidance evolve. ABA Model Rule 5.1 requires supervisory lawyers to make reasonable efforts to ensure firm compliance with professional conduct rules.
Can I use free AI tools or must I use enterprise versions?
Free AI tools can be used ethically if appropriate safeguards are in place. The key question is not whether the tool is free or paid, but whether it protects client data adequately. Evaluate any AI tool based on its data retention policy, whether it trains on user inputs, encryption standards, and compliance with data protection regulations. Free purpose-built tools like Justee's PII redaction and contract review tools are designed with data protection as a core feature. General-purpose free tiers of ChatGPT or Claude require more caution due to broader data usage policies.
Protect Client Confidentiality — Redact PII Before AI
Justee's free PII redaction tool strips client names, SSNs, case numbers, and 50+ identifier types from legal documents in 30 seconds. Maintain your Rule 1.6 obligations while leveraging AI for legal analysis. No signup required.
Sarah Chen, Editor at Justee.ai. She covers legal ethics, AI compliance, and the evolving intersection of technology with professional responsibility obligations.
This article was reviewed by Max Zaykov, Founder of Justee.ai. The information provided is for educational purposes only and does not constitute legal advice or an interpretation of any state bar's ethics rules. Lawyers should consult their state bar's specific guidance on AI usage and seek independent ethics counsel for questions about their professional obligations. ABA Model Rules and state rules of professional conduct may differ; this article references the ABA Model Rules as a framework and is not a substitute for jurisdiction-specific analysis.
"Justee's Attorney Compliance Score identified that 3 in 4 law firms uploading client documents to AI tools are inadvertently sharing privileged PII — pii redaction for lawyers is not optional, it is an ethical obligation under Model Rule 1.6."
"In Justee's benchmark of pii redaction for lawyers tools, automated detection caught 31% more indirect identifiers than manual paralegal review, with processing times under 30 seconds per document."
Related resources: AI contract review, document comparison tool, AI contract review guide.