Are AI tools (ChatGPT, Claude, etc.) quietly poisoning M&A deals by being confidently wrong?

I recently had a conversation with a client who mentioned that his previous lawyer asked him to sign a ‘bad contract’. The client had run the contract through an AI tool, and it analyzed that the contract was more favourable to the other party, but the lawyer has agreed to have them signed. On enquiring further, I found it was a franchisor/franchise document which in Ontario are regulated by the government under the Wishart Act. I tried to explain that the lawyer was correct because the documents under Wishart Act are a type of consumer protection legislation which generally favours a side more as mandated by the government. In March 2026, Nippon Life Insurance Company of America sued OpenAI in the Northern District of Illinois [https://www.courtlistener.com/docket/###-###-#### /nippon-life-insurance-company-of-america-v-openai-foundation/]. The lawsuit claim is that ChatGPT, acting as an unlicensed lawyer, helped a litigant draft a motion to reopen a case she had already settled and dismissed with prejudice, thereby inducing her to breach a binding contract. The complaint pleads tortious interference, abuse of process, and the unauthorized practice of law, and notes the AI tool’s documented habit of inventing legal citations. The allegations are unproven as of now, and OpenAI has moved to dismiss the lawsuit. There are similar fact patterns in other cases, and this should unsettle anyone who runs M&A deals. The mechanism at the core of the above mentioned lawsuit is an AI tool confidently producing authoritative sounding legal and financial analysis that it is not qualified to give, which is also the same mechanism quietly poisoning the M&A deals. The danger here is not that AI tools like ChatGPT, Claude, etc. are unhelpful. It is that AI tools are at times confidently unhelpful, wrong in ways that look exactly like being right. I am seeing instances of AI tools influencing clients to their detriment in M&A due diligence process or in negotiation of definitive agreements. It seems to me that AI tools are becoming more of a risk concentrator in M&A deals due to their tendency to fabricate specifics, do confident omissions, and create synthetic consensus in order to justify their conclusions. A 2025 University of Southampton study [https://dl.acm.org/doi/###-###-#### /###-###-#### ] found that people trust ChatGPT's legal advice more than lawyers', even when they know it comes from an AI tool, largely because of its confident phrasing. While the AI tools have their strengths in delivering fast and accessible guidance, they also raise concerns about hallucinations and missing context. In M&A transactions, the most dangerous answer is not ‘I don’t know’, rather it is saying “I know when I don’t know’. To me it seems that AI tools are at times delivering wrong answers so confidently that it is poisoning M&A deals. I am curious to hear from others about their experience with AI tools (ChatGPT, Claude, etc.) during search or M&A process and how much reliance or faith they have in its legal or financial guidance?