Balfour Capital warns AI profiling can create reputational risk

Balfour Capital Group released a white paper on June 12, 2026 arguing that generative AI can misstate corporate facts, confuse entities and amplify reputational harm if firms rely on it without verification. The report says the risk is especially acute in financial services, where a bad AI summary can affect screening, due diligence and mandates before a firm can correct the record. Why it matters: - Balfour Capital Group says generative AI is now shaping early-stage due diligence for investors, journalists, regulators, compliance teams and prospects. - A false AI summary can affect whether a firm is viewed as licensed, sanctioned or under investigation. - In financial services, that can mean lost opportunities, delayed transactions and higher compliance costs before any human review occurs. What happened: - Balfour Capital Group released a new white paper on June 12, 2026. - The report is titled, “The Reputational Risk of Generative AI: How AI-Generated Corporate Profiling, Hallucination, and Entity Confusion Create Material Risk for Businesses, Investors, and Financial Institutions.” - Steve Alain Lawrence, Balfour Capital Group’s chief investment officer, authored the paper. - The paper argues that AI assistants are increasingly used as the first stop for due diligence and are too often treated as verified sources. The details: - Lawrence says large language models predict the most statistically probable sequence of words, not truth. - The paper says AI output can sound equally authoritative whether it is accurate or fabricated. - The report says AI systems can confuse one company with another entity that shares a similar name. - The paper cites Stanford research showing purpose-built legal AI tools hallucinated on 17% to 33% of queries. - The report also cites an independent legal database tracking more than 700 court decisions worldwide involving fabricated AI-generated citations. - The paper says roughly 90% of those decisions were recorded in 2025. - The report cites cases including Moffatt v. Air Canada, Mata v. Avianca, the Brian Hood matter, the Arve Hjalmar Holmen matter and Deloitte Australia’s AU$440,000 government report found to contain fabricated references. - Balfour Capital says those examples show that unverified AI output has already caused financial, legal and reputational harm. - Lawrence says AI should be treated as a research aid, not a conclusion. - The paper says firms should confirm important facts against primary sources. - The report says trust is especially important in finance because counterparties evaluate firms in stages: preliminary screen, meeting, due diligence and allocation. Between the lines: - The paper is not a rejection of AI. - Its core warning is that speed and fluency can make falsehoods harder to spot than traditional errors. - The report shifts part of the burden from individual users to AI developers and regulators. - Balfour Capital says developers should provide clearer source attribution, confidence signals, better separation of fact and inference, and ways to correct false information about real entities. - The paper also calls for regulators to set transparency expectations around source attribution and data freshness. - The report encourages firms to maintain an authoritative public record of their structure, licensing and history so AI systems have a reliable reference point. What’s next: - The white paper recommends periodic audits of what AI systems say about a firm. - It also calls for plain-language explainers, clear internal ownership and escalation paths, retrievable primary-source documentation and direct engagement with AI developers. - The paper says firms should set advance thresholds for legal action when false AI output causes harm. - Balfour Capital says companies should also educate counterparties that AI is a research aid, not a substitute for due diligence. - The full white paper is available here . The bottom line: - Balfour Capital’s message is simple: AI can accelerate research, but reputation still depends on verified facts and human judgment.