Attorney Ethics & Bar Compliance

A reference brief for attorneys evaluating AI-assisted legal tools · Last updated October 2026

Note: This document is provided for informational purposes only and does not constitute legal advice or an ethics opinion. Attorneys should conduct their own review of applicable Rules of Professional Conduct and consult with their state bar's ethics counsel as appropriate. Rules vary by jurisdiction.

The Core Question

May an attorney use a cloud-based AI tool to assist with drafting, reviewing, and analyzing legal documents without violating duties of confidentiality, competence, or supervision?

The short answer, under the prevailing bar guidance: yes — provided the attorney exercises competent supervision, selects a vendor with reasonable security measures, and understands how the tool processes client data.

Governing Framework

The analysis flows from three Model Rules of Professional Conduct, adopted in substantially similar form by most U.S. jurisdictions:

Rule 1.6 — Confidentiality of Information

An attorney shall not reveal information relating to the representation of a client unless the client gives informed consent or disclosure is impliedly authorized to carry out the representation. Transmitting documents to a cloud vendor for processing is generally considered impliedly authorized under the same reasoning that permits use of email, phone, and third-party copying services — provided the attorney takes reasonable precautions.

Rule 1.1 — Competence

Comment 8 to Rule 1.1 (added in 2012) requires attorneys to maintain competence with "relevant technology." Bar ethics committees have consistently held that this includes understanding how AI tools process data — not building them. An attorney who understands what Ramtek Intelligence sends to its AI inference provider, what that provider does with it, and how to supervise the output meets this standard.

Rule 5.3 — Supervision of Non-Lawyer Assistance

Attorneys must make reasonable efforts to ensure that outside services — including AI tools — operate in a manner compatible with professional obligations. This requires selecting a vendor with appropriate data handling practices and reviewing AI outputs before relying on them. It does not require that AI never be used.

Key Bar Opinions

The two national ABA opinions below apply regardless of jurisdiction — they're the primary authority. Florida and California are included as representative examples of how individual state bars have applied that same reasoning; if your firm is elsewhere, check whether your own state bar has issued comparable guidance (most that have addressed AI tools directly have reached the same conclusion).

ABA Formal Opinion 477R (2017) — Securing Communication of Protected Client Information

Attorneys may use cloud-based tools to transmit and store client information when they apply "reasonable efforts" to ensure security. The opinion explicitly rejects a "perfect security" standard. Factors include: the sensitivity of the information, the likelihood of interception, the attorney's understanding of the technology, and the vendor's security practices.

Conclusion: Cloud AI tools are permissible with reasonable due diligence.

ABA Formal Opinion 512 (2024) — Generative Artificial Intelligence Tools

Directly addresses large language model AI tools in legal practice. Confirms that use of AI is consistent with professional obligations when the attorney: (1) understands the tool's capabilities and limitations; (2) supervises and reviews AI outputs; (3) takes reasonable steps to protect confidential information; and (4) charges only reasonable fees. The opinion does not prohibit AI use — it requires competent, supervised use.

Conclusion: AI tools are ethically permissible. Supervision and competence are the operative duties.

Florida Bar Ethics Opinion 24-1 (2024)

Addresses attorney use of AI tools including generative AI. Florida attorneys may use AI tools that process client information provided they: conduct reasonable due diligence on the vendor's data practices; ensure confidential information is not used to train public AI models; and supervise all AI-generated work product before use. The opinion notes that attorneys are not required to disclose AI use to clients unless the specific circumstances require it under Rule 1.4. It also advises lawyers to consider obtaining a client's informed consent before putting confidential information into a third-party generative AI tool, and says a lawyer who uses a chatbot to talk with prospective clients should tell them they are communicating with an AI program and should not mislead them. Ramtek's Website Intake Assistant labels itself as an AI assistant for that reason.

Conclusion: Firms in Florida can rely on this opinion directly. It's also a useful reference point for firms elsewhere, since it's one of the more detailed state-level opinions specifically addressing generative AI tools.

State Bar of California Practical Guidance (2023, since updated)

California's guidance tracks the ABA framework: attorneys must understand the tool, supervise outputs, protect confidentiality, and ensure the vendor does not use client data for training without consent. The guidance specifically distinguishes between consumer AI products (where data may be used for training) and API-based enterprise tools (where data is not retained or used for training under the provider's commercial API terms).

Conclusion: API-based tools like Ramtek Intelligence fall in the permissible category under California guidance.

How Ramtek Intelligence Satisfies These Standards

Client data is transmitted via a commercial AI inference API — not a consumer product. Per the provider's usage policy, inputs are not used to train AI models.
Each request is stateless — the AI has no memory of prior sessions or of any other firm's data. The AI provider may retain API data for a limited period (up to 30 days) solely for trust-and-safety monitoring, after which it is deleted. It is never used for training.
All data is transmitted using TLS 1.2+ encryption. Documents you draft or analyze are saved to your firm's own account so your team can retrieve them later, are visible only to users at your firm, and are deleted on request, 30 days after your subscription or your free trial ends, or sooner if you ask.
The attorney reviews all AI output before use. Ramtek is a drafting and analysis assistant — not an autonomous actor. Supervisory control remains with the attorney at all times.
Ramtek does not access, share, or sell client data to any third party. Subprocessors are limited to AI inference, application hosting, database hosting, payment processing, email delivery and routing, content delivery and network security, error monitoring, legal research sources, and an internal team-messaging service that receives only the basic details of new sign-ups and trials. The full list is on our Security page. The full list is on our Security page.
Attorneys can request full data deletion at any time by contacting support@ramtekintelligence.com.
Legal research: when you ask a legal question, the AI Law Assistant searches public legal sources, including court opinions in CourtListener and official statute and regulation sites, and may use the AI provider's web search. Search terms are sent to those sources, so do not put client names or confidential facts in a research question. Every case cited is checked against a search result, and the attorney must verify every authority before relying on it.

Practical Guidance for Your Firm

Start with lower-sensitivity documents. NDAs, retainer agreements, standard demand letters, and intake forms carry lower privilege risk than sealed filings or grand jury materials. Build comfort with the tool before using it on your most sensitive matters.
Always review AI output. No AI output should be transmitted to a client or court without attorney review. Treat Ramtek like a first-year associate — capable and fast, but requiring supervision.
Document your due diligence. Keep a copy of this brief and Ramtek's Security & Data Handling page in your firm's vendor files. This demonstrates the competence review required under Rule 1.1.
Consider client disclosure. While most bar opinions do not require disclosure of AI tool use, some clients — particularly sophisticated institutional clients — may prefer to know. Your judgment applies.
Check your jurisdiction. Ethics rules vary by state. If your state bar has issued specific AI guidance, review it. Most adopt the ABA framework; some add jurisdiction-specific requirements. Beyond Florida and California, a long list of bars has weighed in, including the District of Columbia (Opinion 388), New York City, Pennsylvania and Philadelphia, North Carolina, Texas (Opinion 705) and Virginia, and most reach the same conclusion. Many courts and individual judges also require disclosure or certification of AI use in filings, so check the rules of every court you appear in.

Questions for Your Malpractice Carrier

For a fuller file, with the questions to ask any AI vendor, a renewal worksheet and a policy starter, download the AI readiness checklist.

If your carrier asks about AI tool use, the following answers apply to Ramtek Intelligence:

"Does the tool use client data to train AI models?" — No. Our AI provider's API terms prohibit use of API inputs for model training.
"Is data encrypted in transit?" — Yes. TLS 1.2+ on all transmissions.
"Does an attorney review the output?" — Yes. The tool generates drafts and analysis; the attorney reviews and approves all work product.
"Can we get a data processing agreement (DPA)?" — Yes. Contact support@ramtekintelligence.com to request a formal DPA.

References: ABA Formal Opinion 477R (2017); ABA Formal Opinion 512 (2024); Florida Bar Ethics Opinion 24-1 (2024); State Bar of California Practical Guidance for the Use of Generative Artificial Intelligence (2023); ABA Model Rules of Professional Conduct Rules 1.1, 1.6, 5.3. For questions, contact support@ramtekintelligence.com.