AI Due Diligence in M&A: Avoid Blind Spots in Your Deal

AI due diligence is the use of artificial intelligence tools to review a target company’s contracts, records, and data during a merger or acquisition. The tools sort documents, pull out key terms, and flag unusual provisions far faster than a manual first pass. However, they don’t replace legal judgment. Lawyers still have to check the output, decide what matters, and turn findings into deal terms.

Buyers and sellers now ask about these tools early, usually because they want a faster process and a lower bill. Both goals are realistic, within limits. At AEGIS Law, we represent buyers and sellers in middle-market deals, and we use technology where it helps the client. This article explains what AI review does well, where it falls short, the ethics rules that apply, and how to plan a deal around it.

What does AI due diligence actually do in a deal?

Most of the work happens in the data room. Contract review tools read each uploaded document, classify it, and extract terms the deal team asks for. Common targets include change of control clauses, assignment restrictions, exclusivity, most favored nation pricing, termination rights, and indemnities. The output is usually a searchable chart that shows which contracts contain which terms. As a result, the team sees patterns across hundreds of agreements instead of reading them one at a time.

Newer tools built on large language models go further. They can summarize a contract in plain language, answer questions about a group of documents, and draft a first cut of an issues list. For example, a buyer’s lawyer can ask which customer contracts let the customer walk away after a sale. The tool returns a list with citations to the relevant clauses. A lawyer then confirms each one.

Some platforms also help with financial and operational review. They can spot gaps in a document set, match invoices to contracts, or flag missing signatures and expired terms. These checks don’t replace a quality of earnings report, but they help focus it.

Where does AI due diligence fall short?

The tools are only as good as the documents and the questions. If the seller hasn’t uploaded side letters, emails that changed a deal, or oral understandings, no tool will find them. Scanned documents with poor image quality can also trip up text extraction. In addition, the tools can miss a clause that uses unusual wording. They can also flag a clause that looks risky but isn’t, given the rest of the agreement.

Language models add a specific risk: confident errors. A tool may summarize a provision incorrectly or describe a term that isn’t there. That’s why every finding that could affect price, structure, or the purchase agreement needs a human check against the source document. Meanwhile, judgment calls stay with people. A tool can show that forty contracts need consent to assign. It can’t tell you which counterparties will actually object, or whether a consent is worth delaying the closing for.

Finally, AI review doesn’t change what a buyer has to investigate. Litigation, employment practices, tax exposure, permits, and intellectual property ownership still need targeted review and interviews. Our due diligence checklist covers the full scope.

What ethics rules apply when lawyers use AI tools?

The American Bar Association addressed this directly in ABA Formal Opinion 512, issued July 29, 2024. It applies existing rules to generative AI. Lawyers must understand a tool well enough to use it competently. They must protect client confidentiality. They must also supervise the people and tools they rely on, and tell clients what they need to know about how the work is done. ABA opinions aren’t binding on their own, but state bars often follow their reasoning.

Confidentiality matters most in a deal. Data rooms hold trade secrets, customer lists, and employee information. The opinion says that before a lawyer puts client information into a tool that learns from user inputs, the client’s informed consent is generally required. Boilerplate consent in an engagement letter isn’t enough. In practice, that means checking whether a tool keeps or trains on your data, and reading the vendor’s terms before anything is uploaded.

Fees are covered too. Under the opinion, a lawyer billing by the hour can charge only for time actually spent. A lawyer also can’t bill clients for time caused by the lawyer’s own inexperience. For more on how these duties play out in daily practice, see our article on AI in legal practice.

How should buyers and sellers plan for AI-assisted review?

Buyers should raise it at the start. Ask your deal team which tools they’ll use, how client data is protected, and how findings get verified. Agree on the questions the review must answer, because a focused request list produces better output than a general sweep. Then set a budget that reflects the time saved on first-pass review and the time still needed for analysis and negotiation.

Sellers should prepare as if a machine will read every page, because one probably will. Organize the data room, use clear file names, and upload complete, signed, text-searchable copies. Fill gaps before the buyer finds them. A clean set of documents shortens AI due diligence and cuts down on follow-up questions. It also gives buyers less reason to ask for price cuts or special indemnities. Clear disclosure schedules help for the same reason.

Sellers can also use the same tools on themselves before going to market. A self-review of key contracts shows which ones need consent to assign, which have expired, and which are missing signatures. Fixing those gaps early is cheaper than explaining them under deadline pressure. It also lets the seller decide how to present a problem, rather than reacting to the buyer’s version of it.

Both sides should address data handling in the confidentiality agreement. Spell out whether the receiving party may load materials into third-party AI tools. Address how long the data may be kept and how it must be destroyed if the deal dies. These terms are easy to agree on early and hard to fix later.

Frequently asked questions

Does AI due diligence make a deal cheaper?

Often, for the document review part. First-pass contract review is faster, so fewer hours go into sorting and summarizing. Analysis, negotiation, and drafting still take lawyer time. The total savings depend on the size of the document set and how organized it is.

Can a seller stop a buyer from using AI tools on its data room?

A seller can negotiate limits in the confidentiality agreement or data room terms. Common approaches allow only tools that don’t keep or train on uploaded data, or require the seller’s approval of specific vendors. Outright bans are less common and can slow the buyer down.

Who is responsible if the AI tool misses something?

The deal team is. Lawyers must supervise the tools they use and verify important findings. Between buyer and seller, the purchase agreement’s representations, warranties, and indemnities decide who bears the cost of a problem found after closing.

Talk with AEGIS Law

Good technology speeds up a deal, but the results still depend on the questions asked and the judgment applied. AEGIS Law represents buyers and sellers in acquisitions and helps clients run efficient, well-documented diligence. Contact AEGIS Law to talk about your next transaction.

By Scott Levine