Accountant warns that AI financial advice is costing businesses

AI For Business


Canadian businesses that rely on artificial general intelligence tools like ChatGPT for financial, bookkeeping and tax advice are already losing money. Accountants have warned that the problem could lead to bankruptcy by 2026.

That’s according to new research from Dext, which surveyed 500 accountants and bookkeepers across Canada. Half of respondents said they knew of a business that suffered direct financial loss as a result of acting on false or misleading AI-generated advice, including overpayments, missed tax deductions, fines, fines, and compliance issues.

The findings suggest that public AI adoption is accelerating across Canadian businesses, but that misuse of these tools for complex financial decisions poses increasing and costly risks. One expert predicts the risks will only increase as more companies rely on AI artifacts as trusted guidance.

“The damage is no longer a hypothetical,” said Paul Rodder, vice president of accounting product strategy at Dext. “Businesses are already losing money, and accountants are spending valuable time fixing avoidable mistakes, from tax and payroll errors to expense misunderstandings.

“AI plays a powerful role in finance, but there is a fundamental difference between specialized tools built for accounting and bookkeeping and general-purpose chatbots that don’t know a company’s true financial health.”

Use of AI increases as customers challenge experts

Throughout 2025, 76% of accountants and bookkeepers said more clients were using public AI tools to seek financial, tax, and bookkeeping advice. At the same time, 70 percent reported that more clients were using AI-generated output to question or challenge professional advice, and 68 percent reported that more clients suggested that AI could completely replace the need for accounting services.

Related: Dext’s new North American leader in AI, payments and growth

Increased dependence has already led to a spike in errors. Only 7% of respondents said they have never encountered an error caused by public AI advice. 11% said they see such mistakes every day, and 29% said they encounter them weekly.

The most common problem was misinterpretation of business expenses, cited by 44% of respondents, followed by incorrect tax claims or claims at 43%. Other frequently encountered problems include poor personal tax planning, payroll errors, and incorrect business tax planning advice.

Fixing AI mistakes costs hours of work

In addition to direct economic losses, the study also points out that accounting firms are experiencing a decline in productivity, increasing costs for companies.

44% of people who encountered AI-related errors said they spent up to three hours a month fixing mistakes with AI-generated advice. Another 27% spend 4-6 hours and 11% spend 7-10 hours fixing errors.

Risks and regulatory needs in 2026

Looking ahead, accountants predict that risks will increase if companies continue to rely on public AI tools without expert oversight. More than a quarter warn that the risk of bankruptcy or business failure will increase by 2026. Others also pointed to increased misuse of AI output to justify improper or fraudulent claims, increased fines and fines, and increased oversight by the Canada Revenue Agency due to errors and delays in filing.

As concerns grow, so do calls for intervention. 94% of respondents said regulation or restrictions are needed, with two-fifths supporting restrictions on public AI tools that provide financial and tax-related advice, and nearly two-thirds calling for formal regulation.

“As we head into 2026, there could be serious consequences if more companies treat AI outputs as trusted tax and financial advice without expert oversight,” Rodder said. “Going forward, we need to focus on responsible guardrails, clearer limits on financial advice, and better education for businesses about what these tools can and cannot safely use.”



Source link