Due diligence has always required reviewing large volumes of information within tight deadlines. Contracts, financial statements, commercial data, employment records, tax records, and management responses must be considered together to uncover risks and opportunities that affect a transaction’s price and terms. Artificial intelligence enhances the ability to organize and examine this material. It can accelerate repetitive tasks, broaden the scope of analysis, and direct specialists’ attention to relevant issues. This contribution, however, does not eliminate the need for human judgment. The faster the tool, the more important it becomes to clearly define what needs to be investigated, verify the evidence, and interpret its implications.
Adoption has already moved beyond experimentation. In a survey of more than 300 M&A executives released in January 2026, Bain & Company reported that 45% were already using artificial intelligence in their work and that adoption had more than doubled in 2025. Deal sourcing, target screening, and due diligence were among the leading areas of application. In a separate 2025 survey of 1,000 corporate and private equity leaders in the United States, Deloitte found that 35% of those using generative artificial intelligence in M&A employed it in due diligence. These respondents also raised significant concerns, with 67% citing data security and 65% pointing to information quality and availability.
The percentages vary depending on the sample and how usage is defined, but they point in the same direction. The technology is advancing rapidly, while the governance surrounding its use still needs to mature.
In practice, artificial intelligence can classify data room documents, identify missing files, summarize contracts, and locate clauses covering change of control, exclusivity, renewal, termination, and penalties. In financial due diligence, it can help reconcile datasets, compare accounting entries, group transactions, and flag unusual patterns. In commercial due diligence, it can organize customer, product, and regional data to identify concentration, revenue losses, price variations, or changes in behavior. It can also structure questions and answers, link findings to source documents, and prepare exception lists for review. The most significant benefit lies in enabling the team to examine a larger share of the information and devote its time to investigating exceptions, rather than simply producing summaries faster.
These benefits come with limitations that cannot be dismissed as technical details. Models may produce plausible but incorrect conclusions, omit a critical qualification, or interpret a clause without accounting for definitions contained in other documents. Confidential information may also be exposed if the solution lacks adequate controls over data access, storage, and use. The generative artificial intelligence risk profile published by NIST in 2024 highlights issues including confabulation, privacy, information security, and the need for ongoing evaluation. In a transaction, every finding produced with the support of artificial intelligence should be traceable to its source, reviewed by a qualified professional, and protected under the same rules that apply to all other due diligence data.
Human judgment remains central because due diligence does not end when an inconsistency is identified. Its materiality must be assessed, management’s explanation understood, its implications for other areas examined, and a decision made on how to address the risk. A discrepancy may reflect a classification error, a recurring practice, a future obligation, or a sign of business deterioration. Depending on the context, the response may involve adjusting the price, revising projections, requiring a specific indemnity, changing a closing condition, proposing remedial action, or walking away from the transaction. No model takes responsibility for these decisions. Professionals in finance, law, operations, technology, and human resources must bring the evidence together and consider both the available historical information and how the business actually operates.
For sellers, the same analytical capabilities make early preparation even more important. Inconsistencies across contracts, management reports, accounting records, and responses provided to the buyer are likely to surface sooner when large sets of documents can be compared quickly. Before opening the data room, the company should organize its files, reconcile information, document exceptions, review access permissions, and prepare explanations supported by evidence. Artificial intelligence will make due diligence faster and more comprehensive, but the quality of decisions will continue to depend on context, materiality, and accountability. The most effective combination is one in which technology extends the reach of the analysis and preserves time for what cannot be automated: questioning, interpreting, and deciding.