Is Lexis+ AI Safe for Confidential Legal Work?

Is Lexis+ AI safe for confidential legal work? The short verdict

Yes, for most firms, as long as the firm controls how it's used. LexisNexis says it doesn't train its AI models on customer data, and it publishes independent SOC 2 Type 2 reports. Even so, a legal AI tool is only as safe as the contract behind it and the way your lawyers use it.

• Lexis+ AI (made by LexisNexis): fits research-first firms that want generative AI tied to a citator and primary law, from a vendor they already trust with client research.
• Harvey (made by Harvey AI): fits large firms building broad AI programs for drafting, review and workflows. It has the longest certification list of the three.
• OpenLaw for Attorneys (Lexi): fits small and mid-size firms that want an AI paralegal for intake, client follow-ups, documents and billing, not case law research.

One disclosure: we build one of these tools. Lexi, the AI paralegal in OpenLaw for Attorneys, is ours. It gets one entry and one honest take, same as the rest. Lexi is not related to getlexi.io, a different company that also uses the name Lexi.

This page is for law firm owners, partners and administrators who are deciding which AI tools can be trusted with client information. We checked security pages and prices in October 2026.

What does LexisNexis do with your prompts and documents?

LexisNexis says it doesn't use customer data to train or fine-tune its AI models. Prompts and documents are encrypted, and security is checked by outside auditors. That covers the main confidentiality risk named in ABA Formal Opinion 512: a self-learning tool that reuses one client's information somewhere else.

• Training: The LexisNexis Trust Center states: "We do not train or fine-tune our AI models using Customer Data."
• Audits: As of October 2026, the Trust Center lists SOC 2 Type 1 and Type 2, SOC 3 and ISO/IEC 27001, with SOC 2 Type 2 reports published in 2026.
• Encryption: LexisNexis says data at rest is encrypted with AES-256 and data in transit uses TLS 1.2 or higher.
• Pseudonymization: LexisNexis says prompts and user interactions are pseudonymized so it can improve the product. That means some form of interaction data is kept. Ask how long it's kept and whether you can switch that off.
• Subprocessors: LexisNexis keeps a global subprocessor list. Check which model providers handle your prompts and what retention terms apply to them.

What matters is the version you actually signed. Ask for the SOC 2 Type 2 report itself, not a summary, and read the AI terms in your order form. Our guide to SOC 2 compliance for legal tech lists what to look for in the report.

Does using Lexis+ AI waive attorney-client privilege?

No court has held that a firm-directed enterprise legal AI tool like Lexis+ AI waives privilege. The leading case involved a consumer chatbot. In United States v. Heppner, No. 25 Cr. 503 (S.D.N.Y. Feb. 17, 2026), Judge Jed S. Rakoff held that documents a criminal defendant created with the consumer version of Anthropic's Claude were protected by neither attorney-client privilege nor the work product doctrine.

The ruling turned on two facts. First, the consumer privacy policy allowed data retention and disclosure to third parties, so the communications weren't confidential. Second, the defendant used the tool on his own, not at the direction of counsel. Commentators, including Venable, read the opinion to mean that enterprise deployments with contractual bans on training, confidentiality commitments, retention limits and attorney supervision stand on much firmer ground.

ABA Formal Opinion 512 (July 29, 2024) adds the ethics layer under Model Rule 1.6. Lawyers must weigh the risk that information about a representation could be disclosed. They must also get the client's informed consent before entering that information into self-learning tools. The opinion says general boilerplate in an engagement letter isn't enough.

Three habits keep Lexis+ AI on the safe side of both:

1. Use firm-licensed accounts only, never personal or consumer AI logins.
2. Have a lawyer direct and supervise the AI work, and record that in the file.
3. Leave out client identifiers that research doesn't need, and check your state bar's guidance on consent.

How much does Lexis+ AI cost compared to Harvey and Lexi?

None of the three publishes a price list. Every quote goes through sales. Third-party buyer reports put Lexis+ AI well below Harvey per seat, and Harvey reportedly requires a seat minimum. Treat any number below as a starting point for negotiation, not a price.

LexisNexis has no public per-seat rate card for its AI layer. Vaquill reports all-in Lexis+ AI or Protégé seats at roughly $250 to $500 per seat per month, depending on content and firm size. Spellbook's October 2026 roundup of market estimates puts Harvey at about $1,200 to $2,800+ per seat per month, with a minimum of around 20 seats. Harvey doesn't confirm these numbers. OpenLaw for Attorneys (Lexi) also quotes after a demo. Our guide to choosing legal AI explains how to compare quotes across tools that price differently.

Where does Lexis+ AI still win?

Lexis+ AI wins wherever the answer depends on authoritative, current law. Its answers come from LexisNexis primary law, secondary sources and citator data. A general-purpose model can't match that, so Lexis+ AI is the safer research choice when an invented citation would be costly.

• Citation grounding: answers link back to LexisNexis content you can open and Shepardize.
• Vendor track record: most firms already send privileged research queries to LexisNexis, so the relationship and terms are familiar.
• Audit depth: a long list of SOC and ISO attestations is published on a public trust center.

Harvey wins on broad, firm-wide AI programs and custom workflows. Lexi wins on running the firm's day-to-day operations, which neither research tool is built for. For more on research tools, see how AI is transforming legal research.

Can firms use Lexis+ AI and Harvey or Lexi at the same firm?

Yes. Many firms pair a research platform with a separate drafting or operations tool, because the tools do different jobs. Every extra tool brings another vendor contract, data flow and subprocessor list to review, so security review doesn't get lighter.

• Lexis+ AI with Harvey: common at large firms. Reported pricing suggests Harvey costs more with a Lexis integration, so ask both vendors how the data flows between them.
• Lexis+ AI with Lexi: Lexis+ AI handles research, and Lexi handles intake, client communication, documents and billing. The overlap is small. Our page on what Lexi replaces lists the operations tools it consolidates.

For every pair, make sure client data shared between tools stays under enterprise terms. Don't paste it into a consumer chatbot as a shortcut.

Which should you choose: Lexis+ AI, Harvey or Lexi?

Choose based on the job, then check security for each pick separately.

• Choose Lexis+ AI if: your main risk is research accuracy, you already license Lexis+, and you want AI grounded in a citator.
• Choose Harvey if: you're a large firm with budget for a seat minimum and a team to build custom AI workflows.
• Choose OpenLaw for Attorneys (Lexi) if: you're a small or mid-size firm, and the work slowing you down is intake, follow-ups, document collection and billing rather than research.

The overall verdict: Lexis+ AI is safe enough for confidential legal work under its enterprise terms, provided your firm uses licensed accounts, supervises the output and follows ABA Formal Opinion 512.