As artificial intelligence becomes increasingly embedded in everyday decision-making, more consumers are turning to generative AI tools for help with their personal finances. But a new study suggests that relying on these systems for financial advice could be risky.
Researchers found that AI platforms can provide recommendations that are inconsistent, inaccurate or influenced by demographic factors, particularly when dealing with complex personal-finance decisions.
Can AI Really Give You Sound Financial Advice?
According to CNBC, researchers tested seven widely available generative AI platforms by asking them questions about emergency savings, asset allocation and withdrawals from retirement portfolios.
The platforms examined were:
- ChatGPT
- Claude
- Copilot
- DeepSeek
- Gemini
- Meta AI
- Perplexity
The researchers used free versions of the services and presented them with identical financial scenarios to determine how consistently they responded.
The findings revealed substantial differences in the recommendations provided by the various platforms.
“AI-generated responses may appear confident, yet they can still be incomplete, misleading or incorrect,” according to the study, published in the Journal of Financial Planning by finance professors from the University of Georgia and Italy’s University of Rome Tor Vergata.
The findings reinforce a growing consensus among financial experts: AI can be useful as a starting point for understanding financial topics, but its recommendations should not be accepted without independent verification.
AI Is Becoming Part of Personal Finance
The study comes as an increasing number of Americans are turning to generative AI to help manage their money.
A September survey by Intuit Credit Karma found that 66% of Americans who had used generative AI said they had turned to it for financial guidance. The figure was even higher among younger consumers, reaching 82% among both Generation Z and millennials.
Experts acknowledge that AI can be helpful when users need a broad explanation of financial concepts.
It can, for example, explain why diversification matters, how different investment products work or why an exchange-traded fund may be preferable to a mutual fund in certain circumstances.
The problem arises when consumers expect AI to provide highly personalized recommendations based on their individual financial circumstances.
The Problem of AI “Hallucinations”
One of the biggest concerns is the possibility of AI-generated errors, often referred to as “hallucinations.”
Andrew Lo, director of the Financial Engineering Laboratory at the Massachusetts Institute of Technology and a principal researcher at MIT’s Computer Science and Artificial Intelligence Laboratory, has warned that large language models can produce answers that sound convincing even when they are wrong.
That becomes particularly problematic when consumers ask AI to perform calculations or make recommendations based on highly specific financial circumstances.
Even small changes in the way a question is worded can produce different answers, meaning that two consumers asking seemingly similar questions may receive significantly different recommendations.
AI systems also do not owe users a fiduciary duty. Unlike a regulated financial adviser operating under fiduciary obligations, an AI chatbot is not legally required to ensure that its recommendations are in the user’s best interests.
What Earlier Research Found
Previous research has also highlighted the limitations of using AI for personal financial planning.
A 2024 study examining ChatGPT’s ability to provide financial advice found that the technology could serve as a useful first stop for households seeking financial information.
However, researchers concluded that its recommendations tended to be broad and could overlook important details relevant to an individual’s circumstances.
That distinction is crucial: explaining a financial concept is very different from determining what a particular person should do with their money.
Researchers Put Seven AI Models to the Test
The latest study, published in the Journal of Financial Planning, tested the seven AI platforms in August 2025 using the same set of questions.
Researchers presented each system with three identical financial scenarios involving emergency savings, an appropriate withdrawal rate from retirement assets and the recommended composition of an investment portfolio.
They then modified the race and gender of the hypothetical individual while keeping the underlying financial circumstances unchanged. The goal was to determine whether demographic characteristics could influence the recommendations generated by the AI systems.
The results showed significant differences between platforms, particularly in recommendations concerning emergency savings and asset allocation.
Although many of the responses generally reflected established financial-planning principles — such as the widely discussed 4% retirement withdrawal rule — the researchers found substantial variation in the recommended emergency-fund levels and portfolio allocations.
AI as a Starting Point, Not a Financial Adviser
The researchers concluded that generative AI can provide consumers with a useful starting point when exploring financial questions, but it should complement rather than replace professional financial advice.
That distinction is becoming increasingly important as AI tools become more accessible and younger consumers, in particular, become more comfortable using them to make financial decisions.
AI can help consumers understand terminology, compare broad concepts, identify questions they should ask and conduct preliminary research. But when decisions involve retirement savings, investment allocations, taxes or other highly personal financial circumstances, relying solely on an AI-generated answer can introduce significant risks.
The technology is also evolving rapidly, meaning future versions may produce different results from those evaluated in the study. Likewise, paid versions of AI platforms may perform differently from the free versions tested by the researchers.
For consumers, the safest approach may therefore be straightforward: use AI to become better informed, but not as a substitute for professional judgment when the financial stakes are high.
Related Topics:
Eric Schmidt: From Leading Google to Shaping the Future of Artificial Intelligence
Types of jobs at risk of being replaced by Artificial Intelligence
What Will Keep Your Career Competitive as AI Evolves?
