In M&A, clients are increasingly asking about how Generative AI (“GenAI”) can be used to make dealmaking easier, both as a means of analyzing the operations of target businesses and to optimize the deal process. As with any fast-developing technology, GenAI risks go beyond traditional IP and cybersecurity, affecting how we conduct diligence, negotiate terms, and plan deal timelines. Below are a few of the more critical near-term challenges GenAI presents to M&A practitioners, as well as some protective considerations.
Hallucinations in Critical Outputs
Generative systems can be wrong, and often persuasively so. At the time of writing this, using the “free” tier, it took the authors three tries to convince ChatGPT that there are, in fact, three Rs in the word “strawberry” and not just one or two. By contrast, ChatGPT identified three Rs when using the “pro” tier, indicating troubling variances in accuracy depending on one’s GenAI subscription. The potential risk of inaccuracy (called “hallucinations” in GenAI-parlance) is vital to remember when using GenAI to help you conduct or produce diligence, draft deal documents, and administer a deal, and when assessing and planning for the risks of buying a business.
A. Use by M&A Lawyers. Many law firms are using sophisticated GenAI tools that are tailored to legal practice to help them serve their clients more efficiently and, hopefully, more effectively. Services like Harvey (which is a GenAI tool similar to ChatGPT, but closed-system), Dealcloser (a transaction management system), and Kira (a diligence tool) can save time and money and even increase accuracy. They have outputs more closely calibrated to what lawyers would expect to see than general tools like ChatGPT or Gemini can provide. However, even these highly-tuned systems make mistakes – especially depending on the prompts and data they are fed.
Accordingly, you can protect yourself with a few simple rules.
- Adopt and enforce a GenAI policy based on the advice of specialists in this area.
- Set clear guidelines for the use of GenAI throughout the project and track usage.
- Require training to prevent disclosure of confidential information.
- Review all outputs as though they are the work of a first-year associate who is doing their best but makes mistakes.
B. Use by Targets. When GenAI is used in the administration or operations of a business, hallucinations can also create liabilities the buyer may inherit. Courts have enforced AI-made promises and misinformation as company commitments. In 2024, the British Columbia Civil Resolution Tribunal decided that Air Canada had to honor a nonexistent “bereavement fare” its chatbot invented when asked about policy by a customer, which resulted in the airline having to pay that customer $812.02 in damages and tribunal fees—illustrating apparent-authority, contract, and consumer‑protection risks that won’t vanish at closing. Judges have sanctioned lawyers for briefs with AI‑fabricated case citations. Mata v. Avianca, Inc., 2023 WL 4114965 at *1 (S.D.N.Y. June 22, 2023) is among the most well-known, but these sanctions are being issued in the United States and abroad in increasing number. Further, public sector and enterprise chatbots have dispensed advice that was actually illegal (such as denying anti-discrimination obligations of landlords in New York City), which could expose the target (and post‑closing, the buyer) to sanctions, investigations, and follow‑on litigation.
Hallucinations can and do have real-world consequences beyond those to the user. In 2023, Google’s parent company’s shares dropped $100 billion after its new AI chatbot made embarrassing errors. In late 2024, Deloitte had to refund the Australian government after delivering a $290,000 report to the country’s Department of Employment and Workplace Relations that included AI‑fabricated citations. Though GenAI is still relatively new, headline-grabbing calamities resulting from bad AI outputs have been plentiful enough that we should be wary of these leaking into disclosures, forecasts, and deliverables, forcing restatements, refunds, clawbacks, or goodwill impairments after a deal closes.
Integration – The Value of Consulting Specialists
GenAI has already been integrated into many companies from top to bottom and inside and out. Unsurprisingly, lawyers are playing Whack-A-Mole with the legal challenges popping up. M&A lawyers have to be great at issue-spotting generally, so they can lean on specialists to make sure they identify red flags across diligence and build out appropriate representations and warranties or line item indemnities to protect their clients. A few areas where GenAI has caused unforeseen (or at least unintentional) liability involving specialists include:
A. Information Technology and Intellectual Property. Lawyers in these technology subspecialities are deepest in the trenches on GenAI. Consult your in-house or external counsel IP and IT attorneys about developing case law and statutes regarding output ownership, including novel questions regarding copyrights and other protections. Make sure to flag any contracts that might include IP licenses or development of products or IP that uses GenAI. Experienced counsel can help ensure the buyer will actually own or have rights to what it thinks it is purchasing.
B. Data Privacy. GenAI can pose risks not only due to the data that companies are inputting into systems, but also because bad actors are increasingly leveraging GenAI as an aid to cyberattacks. In November 2025, Anthropic (developer of the Claude family of models) released a report that it had detected a “highly sophisticated cyber espionage operation … target[ing] roughly 30 entities” and demonstrating “unprecedented integration and autonomy of AI throughout the attack lifecycle.” Collaborate with your data privacy specialists to determine how well-protected a target’s systems are and find ways to secure the assets you are purchasing as you draft the acquisition agreements.
C. Labor and Employment. Consult labor and employment specialists to plan for discrimination risks when employers use generative AI in hiring and management, because laws like Title VII, the ADA, and the ADEA apply to algorithmic tools. There is also a proliferating patchwork of state and local regimes to consider, including NYC Local Law 144, and Colorado and Illinois’ state-level laws. In Mobley v. Workday, Inc., 2025 WL 1424347 (N.D. Cal. May 16, 2025), the plaintiffs alleged that Workday’s AI software played a role in age discrimination and the court asserted, “Nothing in the language of the federal anti-discrimination statutes or the case law interpreting those statutes distinguishes between delegating functions to an automated agent versus a live human one.” Also, consider how privacy issues overlap with labor laws, wage/hour reclassification issues as duties shift with the use of GenAI, and other issues these lawyers can help analyze.
Rest assured, the challenges associated with GenAI will continue to evolve as rapidly as the technology itself. By staying diligent, you can make GenAI work for you, managing hallucinations and integrating specialist review with your use of this amazing technology. There may be hallucinations in GenAI, but its value is not a mirage – you can integrate this technology into your practice with great success if you stay diligent.
This newsletter may be reproduced, in whole or in part, with the prior permission of Thompson Hine LLP and acknowledgement of its source and copyright. This publication is intended to inform clients about legal matters of current interest. It is not intended as legal advice. Readers should not act upon the information contained in it without professional counsel.
This document may be considered attorney advertising in some jurisdictions.
© 2026 THOMPSON HINE LLP. ALL RIGHTS RESERVED.