AI and E-Discovery in Florida: Privilege, Proportionality, and Rule 1.280

Using AI to review discovery in Florida litigation raises privilege and proportionality questions under Rule 1.280. Here is how to deploy AI document review without waiving privilege or violating the discovery rules.

What Florida Rule 1.280 Actually Requires

Florida Rule of Civil Procedure 1.280, "General Provisions Governing Discovery," defines the scope and limits of discovery in civil litigation. For attorneys considering AI-powered document review, two sections are critical: 1.280(b)(1) on scope and proportionality, and 1.280(b)(6) on privilege. The rule, effective as amended, requires that discovery be relevant to the subject matter and proportional to the case, considering factors like the amount in controversy and the parties' resources. It also mandates that parties take reasonable steps to prevent the disclosure of privileged information. Failure to adhere to these requirements can lead to sanctions, waiver of privilege, and malpractice claims.

The core challenge is applying a 20th-century rule to 21st-century technology. The volume of electronically stored information (ESI) in a typical commercial or personal injury case can run into terabytes. Manual review is no longer feasible, making AI not a luxury but a necessity for meeting the proportionality demands of Rule 1.280.

Proportionality and AI: Meeting Rule 1.280(b)(1) Obligations

Rule 1.280(b)(1) explicitly limits discovery to that which is "proportional to the needs of the case." The court considers the importance of the issues, the amount in controversy, the parties’ access to information, the parties’ resources, and the importance of the discovery in resolving the issues. In a world of exponential data growth, AI-assisted review is one of the most effective tools for satisfying this mandate.

Consider a 12-attorney commercial real estate firm in Tampa representing a developer in a dispute over a failed multi-use project. The ESI involves years of emails, architectural plans in various formats, financial projections, and municipal zoning communications, totaling over two million documents. A manual review by associates and paralegals would take months and cost hundreds of thousands of dollars, a burden that may be disproportionate to the $1.5 million at stake.

By using an AI document review tool, the firm can rapidly categorize this universe of documents. The system can be trained on a small set of exemplar documents (e.g., communications with the primary lender) to identify and prioritize similar documents for human review. This process, often called Technology-Assisted Review (TAR) or predictive coding, reduces the volume of non-relevant documents an attorney must see by over 90%. This directly addresses the proportionality factors in Rule 1.280(b)(1) by drastically reducing the time and cost (the "burden or expense") of discovery.

This isn't just theory. Courts are increasingly acknowledging AI's role in proportionality. While Florida courts have not issued a landmark ruling mandating TAR, the federal trend, beginning with cases like *Da Silva Moore v. Publicis Groupe*, has been to accept and even encourage its use to manage discovery costs.

Privilege and Confidentiality: AI Under Rule 1.280(b)(6) and ABA Rule 1.6

The most significant compliance risk in using AI for discovery is the inadvertent disclosure of privileged information, governed by Rule 1.280(b)(6) and the foundational duty of confidentiality under Florida Rule 4-1.6. Sending client data to a third-party cloud AI vendor for processing creates a serious risk.

ABA Model Rule 1.6, which Florida's Rule 4-1.6 mirrors, requires a lawyer to make "reasonable efforts to prevent the inadvertent or unauthorized disclosure of, or unauthorized access to, information relating to the representation of a client." The key term is "reasonable efforts." Uploading thousands of confidential client documents to a cloud service whose data handling, security protocols, and employee access policies are opaque may not meet that standard.

A recent formal opinion from the New York State Bar Association, NYSBA Ethics Opinion 1280 (May 2024), analyzed the use of generative AI and emphasized that lawyers must understand the technology and ensure confidentiality is protected. The opinion cautions against using AI platforms that may use client data to train their models, as this could constitute a breach of confidentiality.

Imagine an 8-attorney employment law practice in Jacksonville handling a wrongful termination case. The discovery includes sensitive employee medical information, internal HR investigation notes protected by work product, and attorney-client communications. Using a cloud-based AI tool like CoCounsel, which costs between $500 and $1,500 per month, requires uploading this entire data set to their servers. This action places confidential information outside the firm’s direct control, creating a potential violation of Rule 4-1.6. If that vendor suffers a data breach, the firm may be held responsible for the disclosure.

Furthermore, under ABA Model Rule 5.3 (Supervision of Nonlawyer Assistance), attorneys have a duty to supervise nonlawyers, which extends to technology vendors. This means the firm must conduct due diligence on the vendor’s security and confidentiality measures. For many small and mid-sized firms, performing this level of technical vendor vetting is impractical.

A Compliance Checklist for AI in Florida Discovery

To ensure compliance, firms should adopt a clear, documented process for vetting and using AI tools in discovery. This table provides a starting point.

Rule / Obligation	Requirement	AI Compliance Action
FL Rule 1.280(b)(1)	Discovery must be proportional to the case.	Use AI to cull non-relevant documents, reducing review time and cost. Document the process and cost savings to justify proportionality.
FL Rule 4-1.1 & Comment	Maintain competence, including understanding the benefits and risks of relevant technology.	The firm must understand how the AI tool works, its limitations, and its data handling policies before using it on client matters.
FL Rule 4-1.6	Make reasonable efforts to prevent inadvertent disclosure of confidential client information.	Prioritize on-premise, local AI solutions that keep client data within the firm’s network. Avoid cloud tools for sensitive ESI.
FL Rule 4-5.3	Supervise nonlawyer assistants, including technology vendors.	If a cloud vendor is used, conduct thorough due diligence on their security, data privacy, and access controls. Secure a written agreement confirming data will not be used for training.
FL Rule 1.280(b)(6)	Protect privileged communications and work product.	Use AI to run privilege screens and keyword searches, but ensure final privilege calls are made by a qualified attorney. Implement a clawback agreement (Rule 1.280(b)(6)(B)).

What Mi Assist Legal Does

Our team at Mi Assist Legal provides an on-premise AI document search and review solution, OpenClaw Legal, that directly addresses these compliance challenges. The system runs entirely inside your firm’s network on a dedicated Mac Mini or Docker host. We use open-source models like Llama 3.2 via Ollama and index your case files locally with ChromaDB. At no point does your client data leave your physical control or touch a third-party cloud server. This architecture is designed to provide the efficiency of AI review while maintaining strict compliance with your duties under Florida Rule 4-1.6.

Frequently Asked Questions

How does an on-premise AI tool handle large document sets without the power of the cloud?

Modern local hardware, like a Mac Mini with an M-series chip, is powerful enough to index and analyze millions of documents efficiently. For a typical case file under 500,000 documents, performance is nearly instantaneous. The system is optimized to run locally, avoiding the latency and security risks of cloud transfers.

Is using an open-source model like Llama 3.2 compliant and secure?

Yes. The key is *how* the model is run. OpenClaw Legal runs these models locally using Ollama. The model operates on your server and has no connection to the internet or its original developers. Your data is never used to train the model, completely isolating your client information.

Do we still need human attorneys to review documents if we use AI?

Absolutely. AI is a tool to assist, not replace, attorney judgment. It dramatically reduces the number of irrelevant documents a lawyer needs to see, flags potentially privileged documents, and surfaces key information. However, the final determination on relevance, privilege, and production always rests with a licensed attorney, in accordance with your duties under Florida Rule 4-1.1 (Competence).

What is the setup process like for OpenClaw Legal?

Our team handles the entire installation and configuration, which typically takes less than a day. We install the hardware in your office, connect it to your network, and work with your IT staff or MSP to securely index the case files you designate. We then provide training for your attorneys and paralegals.

Our team can help your firm evaluate its discovery workflow and determine if an on-premise AI solution is the right fit. We offer a 4-week sandbox pilot to demonstrate the capabilities of OpenClaw Legal using your own data, securely within your own office.

Schedule a 30-minute on-site assessment with our team to explore a compliant path for AI-powered discovery.