AI and accountants: The rules and guidelines you need to follow

As Making Tax Digital increases reporting frequency and client touchpoints, firms need a scalable way to manage growing workload.

At the same time, we’re entering a new phase of AI in accounting.

The two make for a powerful combination, but there might be reservations given that accountants handle so much client data, and invite so much trust.

Luckily, the Professional Conduct in Relation to Taxation (PCRT) guidance from seven accountancy and tax bodies has been updated to explicitly discuss deploying AI.

In this article, we take a look and also explore how agentic AI systems are reshaping accounting workload. We also set out a clear, practical roadmap for embedding it into your practice.

Here’s what we discuss.

New PCRT guidance on using AI in tax

In January 2026, the UK professional bodies behind the PCRT published new guidance on the use of AI in tax work.

The guidance doesn’t attempt to explain how to implement AI tools. Instead, it interprets the five fundamental principles already present in the PCRT, but through an AI lens.

PCRT is required reading in any event if you’re an accountant, bookkeeper, or tax adviser. But the core messages when it comes to AI are as follows:

• Integrity: Be transparent about how AI supports work and ensure outputs are accurate and honest. An example is ensuring a clear understanding of how an AI-assisted conclusion was reached, so you can explain and justify it confidently if a client queries it.
• Objectivity: Avoid over-reliance. Challenge outputs. Guard against bias. An example might be always critically assessing an AI-suggested tax treatment rather than accepting it without considering the client’s specific circumstances.
• Professional competence and due care: AI does not replace expertise. Understand its limitations and review outputs carefully. Examples include checking that an AI-generated calculation reflects current tax rates and relevant thresholds before submission.
• Confidentiality: Protect client data, particularly when using external or uncontrolled systems. This could include only uploading anonymised client financial information into publicly accessible AI tools, to avoid breaking data protection laws.
• Professional behaviour: Maintain the profession’s reputation and comply with regulatory expectations. For example, you should carefully review any AI-drafted correspondence to HMRC to ensure the tone and professionalism reflect your firm’s standards.

The underlying principle is simple: firms and individual members remain fully accountable for the accuracy, appropriateness, and integrity of their work, regardless of whether that work is AI-created, AI-assisted, or otherwise.

Frank Haskew, Head of Taxation Strategy at ICAEW, adds that failure to apply these principles has consequences:

“If a member fails to adhere to the principles set out in PCRT they are liable to be subject to the disciplinary process. We have seen examples of recent tax cases where [the guidance] has not been applied—for example, using AI to generate statements and case details which are not supported by the facts.”

Integrated AI in modern practice

AI in accounting practice is no longer about automating simple tasks like drafting an email response.

It now works more like a permanent member of the team: analysing financial data, flagging issues, and supporting decision-making, so you can focus on higher, more sophisticated service levels, or simply acquiring new clients.

And as AI becomes more integrated into tax workflows, it serves as a reminder that protecting client data is nonnegotiable. The updated PCRT guidance makes it clear that confidentiality remains a fundamental professional obligation.

This means accountants are turning to secure, closed AI environments designed specifically for financial information, rather than using open public tools where client data might be ingested without their knowledge to train future AI models.

Client personal information or data ending-up in a large language model (LLM) accessible by basically anybody worldwide is a very real risk.

With products like Sage Copilot and the MTD Agent in Sage for Accountants, AI operates inside the accounting software itself.

The Sage MTD Agent is completely free for Sage Accountants—and it’s built to cut compliance workload in half. From segmenting your client base and building task lists to sending reminders and chasing missing documents, it handles the repetitive legwork that eats into your day. Watch this video to get an idea of how it can help you.

Smarter review across every client

AI accounting assistants analyse client data in real time, helping you manage increasing reporting demands by identifying what needs review. This includes the ability to:

• Detect patterns in income volatility that could affect payments on account
• Flag missing transaction data before submission periods
• Identify clients approaching VAT or other tax thresholds
• Flag overdue invoices by value and contact
• Highlight reconciliation discrepancies in real time
• Generate structured quarterly summaries ready for professional review
• Compile relevant HMRC guidance linked to flagged issues
• Compile client tax position overviews
• Analyse large datasets to support client due diligence and AML processes

When you’re a small practice juggling compliance across dozens of clients, your time is finite. AI assistants help direct attention to the areas that warrant closer review.

It reduces your time spent searching for issues and increases the time you can dedicate to informed, client-focused advice.

The next evolution is agentic AI systems that do more than respond to prompts. The aforementioned Sage MTD Agent is a great example.

Agentic AI monitors client data continuously, identifying changes and flagging risks as they arise. Soon, they will become the main way we all interact with accounting software.

For accounting tasks, agentic AI features include:

• Continuously monitoring cash flow trends
• Proactively alerting you to emerging risk patterns
• Triggering follow-up prompts when tax thresholds are met
• Suggesting advisory opportunities as financial behaviour changes
• Initiating workflow actions based on predefined rules

So instead of logging in to check whether something needs attention, agentic AI will tell you.

In the new era of MTD for Income Tax, with quarterly submissions layered on top of existing VAT obligations, this proactive capability is becoming essential. As reporting frequency increases, so does the pressure on time.

Continuous monitoring helps absorb that pressure without extending your working hours.

A practical guide to starting your AI journey in practice

Adopting AI doesn’t require a complex change programme or a large team—just clear objectives, defined ownership and a practical plan.

Here’s a clear, step-by-step approach.

1. Define the outcome

Start with the pressure points in your own firm. Are you looking to:

• Manage the increased workload created by MTD?
• Improve turn around times?
• Free up time to win new clients?
• Focus on growth and planning conversations with clients?

Clear goals ensure AI supports your workload instead of complicating it.

Once your objective is defined, translate it into measurable milestones. For example:

• 50% of clients supported by AI-assisted workflows by the end of 2026
• A defined reduction in manual processing or review time
• Earlier identification of anomalies across quarterly submissions

Clear metrics create accountability and help ensure AI delivers real impact quickly.

2. Choose technology that fits your ecosystem

With goals in place, invest in technology that strengthens your current processes and enables your next stage of growth.

Prioritise AI assistants that work within secure accounting systems rather than disconnected public tools. These integrated solutions:

• Surface issues without additional manual checks
• Monitor deadlines in the background
• Protect confidential client data
• Align with professional standards

3. Start small and build confidence

There is no need to roll AI out across every client immediately.

Begin with your own practice accounts. Apply intelligent workflows to your bookkeeping, VAT returns and reporting processes. This provides hands-on understanding in a low-risk environment.

Once you’re confident, extend AI-assisted workflows to selected client engagements—particularly those with heavier compliance demands.

Monitor how it improves review time, exception detection and client responsiveness. Adoption is more likely to stick when you can point to tangible gains.

4. Establish ownership and internal standards

Clear ownership is critical. If your firm has more than one partner, appoint one to take responsibility and lead the rollout. The role doesn’t need to be technical. It’s about testing features, sharing wins, monitoring impact and keeping the momentum moving.

Alongside ownership, create simple internal guidelines. This doesn’t need to be a lengthy policy, but it should be clear:

• Which AI tools are approved
• What data can and cannot be entered into them
• How AI-assisted work is documented to keep a proper audit trail
• That all AI-generated output must be reviewed before it is sent externally

If you have a team of accountants, it’s important to train them not only how to use AI tools effectively, but how to apply professional judgement when reviewing outputs.

5. Embed transparency and safeguards from the outset

Using AI responsibly includes being transparent with clients about how it supports your work.

Consider how you communicate the use of AI tools to clients. For example, you could include an appropriate statement in engagement letters, indicating that AI-enabled software may be used in delivering services.

You should also review your data protection processes, including GDPR and the Data Protection Act. Your policies should reflect how AI tools are used and how client information is protected.

Importantly, support is available to help navigate AI adoption.

Final thoughts on AI and accountants

AI in accounting is about incorporating intelligent systems that extend your firm’s capability.

The updated PCRT guidance provides the professional framework for doing this confidently and responsibly.

The firms that thrive will not be those that adopt AI fastest, but those that embed it strategically.

When integrated within established standards, AI moves beyond simple automation to become reliable, structured intelligence that strengthens decision-making.