IRS AI Guidance: What Tax Professionals Need to Know
AI can help tax professionals work faster, but it doesn’t replace your professional judgment, responsibility or obligation to protect client information.
In this episode of the Financially Fit Business podcast, I talk with Les Bryson, founder and CEO of The Lab Business Solutions and a 20-year tax consulting veteran, about IRS guidance addressing the use of artificial intelligence in federal tax practice.
This is Part 1 of our three-part series. We start with some of the fundamentals: professional competence, reviewing and verifying AI-generated work, creating a documented process for your firm, and protecting sensitive client data.
If you or your team are already using AI in tax or accounting work, this is a conversation worth having now.
AI Doesn’t Replace Your Professional Expertise
One of the most important points Les makes is also one of the simplest: AI should help you do your work more efficiently. It shouldn’t become the expert you’re relying on to make professional decisions.
If you’re using AI for tax work, you need to understand the subject matter well enough to recognize when the information it gives you is wrong.
That means understanding both your area of tax expertise and the limitations of the AI tool you’re using.
Understand the AI Tools You’re Using
You don’t have to become a software engineer to use AI responsibly. You do, however, need to understand enough about the technology to recognize its limitations.
AI systems can produce inaccurate information. The way you structure a prompt can also influence the response you receive. Les emphasizes the importance of understanding how the tool works, where its blind spots may exist and what you need to verify before relying on its output.
That’s especially important when the information will become part of work delivered to a client.
Review Everything AI Helps Create
Due diligence doesn’t disappear because AI helped produce the work.
Les’s advice is straightforward: any work going out to a client should be reviewed by someone who understands the subject matter, whether the initial work was completed by an employee, an intern or an AI system.
That means checking facts, tax code references, calculations, conclusions and other information before the work leaves your firm.
Speed is one of AI’s biggest benefits, but speed shouldn’t come at the expense of accuracy.
Create a Written AI Review Process
Individual review is important, but firms also need a repeatable process.
Les recommends establishing a written workflow for how AI-assisted work will be reviewed. That process can identify who’s responsible for each step, what needs to be checked and where final responsibility rests.
You may not need to create an entirely new workflow. Many accounting firms already use checklists and defined review processes. AI-related verification can become another part of those existing systems.
The important thing is consistency. Your team should know what’s expected every time AI is used in client work.
Protect Client Data When Using AI
Client data adds another important consideration.
Les discusses the differences between free, paid and enterprise-level AI tools, including differences in data retention and whether information may be used for training.
But even stronger privacy settings shouldn’t become an excuse to put sensitive information into an AI system unnecessarily.
His practical recommendation is to scrub the data first. Remove client names, identifying information and other sensitive information before entering material into an AI tool.
Know What You’re Agreeing To
It’s also important to understand the terms and settings associated with the AI tools your firm uses.
What happens to the information you enter? Is it retained? Can it be used for training? What privacy and security controls are included with your particular account?
Those terms and capabilities can differ depending on the product and subscription level, and they can change over time.
Don’t assume that because you’re paying for an AI tool, every data protection setting you need is automatically enabled.
Build AI Into Your Existing Professional Standards
The takeaway from this first conversation isn’t that tax professionals shouldn’t use AI. It’s that using AI doesn’t remove the professional standards that already apply to your work.
You still need subject-matter expertise. You still need to review the work. You still need a reliable process. And you still need to protect your client’s information.
AI can make your firm more efficient, but you remain responsible for what ultimately goes to the client.
Listen to my full conversation with Les for more about building a responsible process for using AI in tax and accounting work.
Questions Tax Professionals Are Asking About AI
Can tax professionals use AI in their work?
The issue isn’t simply whether you use AI. The discussion in this episode focuses on how you use it while continuing to meet your professional responsibilities. You need the subject-matter expertise to evaluate AI-generated information, review the work before it reaches a client and protect sensitive client information.
Do I need to verify information generated by AI?
Yes. Les emphasizes that AI shouldn’t replace your expertise. If you’re using AI for tax work, you need to understand the subject well enough to recognize errors and thoroughly review the resulting work.
Should an accounting firm have a written AI policy or process?
Les recommends a documented process for reviewing AI-assisted work. That can include a checklist, defined review steps and clear responsibility for who verifies the work before it’s delivered to a client.
Should I enter client information into an AI tool?
Les recommends removing client names, identifying information and other sensitive data before putting information into an AI system. He also recommends understanding the data retention, training and privacy settings associated with the particular AI service you’re using.
Are paid or enterprise AI tools safer for client data?
Different subscription levels can provide different privacy and data-retention options. Les’s recommendation is to understand exactly what your particular service provides rather than assuming that a paid account automatically provides all of the protections your firm needs.