Artificial intelligence (AI) is no longer just an emerging technology; it has become a staple in many workplaces. By February 2026, a Gallup survey revealed that over half of U.S. employees and 68% of those in professional services were utilizing AI in their work. Similarly, a Thomson Reuters report from spring 2026 stated that 74% of professionals engaged AI multiple times a week.
The benefits of AI in professional settings like accounting, law, and consultancy are manifold, ranging from improved research capabilities to enhanced communication efficiency. However, the swift adoption of AI has introduced significant risks, particularly when AI generates incorrect information that can appear both polished and authoritative, as noted by CNN. This raises the crucial issue of accountability, as professionals, not AI developers, bear the responsibility for AI-induced errors.
Misleading Citations
In the realm of tax accounting, a noteworthy example is the Clinco v. Commissioner case, analyzed by two certified public accountants in the Journal of Accountancy’s September 2026 issue. The case involved an attorney presenting a brief with purportedly fabricated citations. The U.S. Tax Court, a specialized body addressing taxpayer disputes, found that three of the four citations used by the taxpayer’s attorney, Abraham Wagner, were AI-generated hallucinations, referencing nonexistent cases. Although Wagner was not sanctioned, the court’s decision serves as a warning against AI-fabricated citations.
iStock/Getty Images Plus
International AI Challenges
AI-related errors are not confined to legal briefs or the United States. In 2025, Deloitte faced issues when a report they prepared for the Australian government included fabricated citations and a false court judgment quotation. After identifying these errors, attributed to generative AI, Deloitte issued a revised report and partially refunded the fees.
Similarly, in Canada, an inaccurate response by Air Canada’s chatbot regarding a bereavement policy led to a legal dispute. The British Columbia Civil Resolution Tribunal ruled against the airline, resulting in damages awarded to the customer.
AI and Legal Ramifications
Legal challenges involving AI platforms continue to surface. In 2025, a defamation case against OpenAI in Georgia was dismissed despite its tool erroneously accusing a radio host of embezzlement. The court found no established fault or damages. In a similar ongoing case, Keene v. Google LLC, a TV producer claims Google’s AI summaries contained false information about his criminal history. The court has allowed some defamation claims to proceed.
Understanding AI Errors
AI errors, often termed “confabulations,” can present incorrect information with plausible explanations and citations, making them difficult to identify without independent verification. The National Institute of Standards and Technology’s 2024 report highlights these issues. Researchers differentiate between hallucinations, where AI invents data, and confabulations, where AI distorts real information in convincing ways.
Addressing AI Mistakes in Professional Settings
A Stanford University study in 2024 highlighted that AI tools for legal research produced erroneous results in up to a third of cases tested. Despite this, many law firms continue to employ AI for its efficiency benefits. However, a 2026 Nationwide survey revealed that only a minority of businesses have formal AI policies or provide training for responsible AI use.
It is crucial for businesses and professionals to establish clear guidelines on AI use, determine which outputs need verification, and ensure that accountability for final work remains intact. AI may transform the professional landscape, but responsibility remains unchanged.
