Building a Firm AI Policy That Supports Responsible Innovation
It’s 9 p.m. on Friday, and a client report is due Monday. You still have substantial data to analyze and documents to review. A new technology tool promises to cut that work in half. There is one complication: your firm requires approval before you can use it, and you worry that the review could take weeks.
As you weigh your options, the manager calls for a status update.
This is the kind of dilemma firms should anticipate when developing policies for artificial intelligence and other technology tools. Professionals need clear expectations and a timely process for evaluating new ideas. An effective policy should help them improve how work gets done while protecting client information and the quality of the firm’s services.

Start with Professional Responsibilities
For firms with an accounting and auditing practice, the AICPA Statements on Quality Management Standards (SQMSs) provide the foundation. SQMS No. 1 requires a system that gives the firm reasonable assurance that the firm and its personnel meet professional and applicable legal and regulatory responsibilities and that engagement reports are appropriate. Ultimate responsibility and accountability rest with the firm’s chief executive, managing partner, or appropriate managing board.
Systems were required to be designed and implemented by Dec. 15, 2025. The standard’s risk-based approach calls for firms to identify and respond to quality risks, including those arising from AI. Governance and leadership, technological resources, engagement performance, and monitoring all have implications for technology policy. In practice, this means assessing tools for their intended uses, training personnel, and providing appropriate supervision and review.
Tax practice has its own requirements. For AICPA members providing tax services, Statement on Standards for Tax Services No. 1, Section 1.4, permits reasonable reliance on tools, including AI, while requiring professional judgment and care. Members remain responsible for the completed work and should take reasonable steps to determine whether a tool is appropriate for its intended purpose.
Practitioners subject to Circular 230 must also meet its due diligence requirements. The IRS’s June 2026 AI guidance reinforces the need to verify AI-generated facts, citations, and calculations before relying on them.
Standalone consulting engagements generally fall outside the SQMSs’ scope. They remain subject to applicable consulting standards and the AICPA Code’s General Standards Rule, including professional competence, due professional care, planning and supervision, and sufficient relevant data. Firms should consider consistent safeguards across service lines, tailored to each engagement’s requirements and risks.
Keep Pace with Legal and Regulatory Developments
Privacy, data security, and client confidentiality should be central to every technology policy. Firms also need to monitor state requirements governing AI in licensed professions and disclosures when clients interact with AI. The applicable obligations depend on the jurisdiction and the activity involved.
Billing is another developing issue, as the IRS Office of Professional Responsibility’s June 2026 introductory guidelines address how AI-related cost savings should be reflected in client fees. Firms should distinguish binding requirements from evolving guidance and involve legal counsel when interpreting their obligations.
Make the Policy Practical
A useful policy should explain how employees can propose improvements and what happens next. Six elements can help:
- Bring the right perspectives together. Include practitioners, technology and cybersecurity professionals, quality management leaders, and legal advisers as needed. Staff and seniors can identify recurring inefficiencies and help assess whether proposed solutions will work in practice. Smaller firms can combine responsibilities and use outside expertise.
- Define what approval covers. Specify the permitted uses, data, account types, settings, and review requirements for each tool. Approval for drafting general communications should not automatically authorize uploading confidential client records or using the tool to support an engagement conclusion. Review how the provider stores, shares, and uses submitted data, including any use for model training.
- Establish a timely review process. Give employees a clear submission channel and an expected time for an initial response. Match the depth of review to the proposed use and its risks. A prompt response might identify an approved alternative, authorize a controlled evaluation with appropriate test data, or explain what further due diligence is needed.
- Set expectations for validation and oversight. Evaluate whether the tool performs reliably for its intended purpose and define the human review required. Establish a process to address inaccurate outputs, unsupported conclusions, and fabricated citations. Clarify who reviews the work, what documentation is retained, and when concerns should be escalated.
- Make the policy easy to follow. Maintain an accessible list of approved tools and uses, provide practical training, and explain the reasons for restrictions. Employees should know where to ask questions, how to promptly report errors or unintended disclosures, and how to make recommendations.
- Review and improve. Revisit the policy as technology, vendor terms, professional standards, and legal requirements change. Reassess tools when their capabilities or intended uses change. Recognize employees whose suggestions produce measurable improvements in quality, efficiency, or client service through the approved process.
Return to the Friday Night Dilemma
When the manager calls, you explain the remaining work and propose the new tool. She directs the suggestion to the firm’s designated technology contact. The initial assessment indicates that due diligence will extend beyond Monday’s deadline.
Your manager helps adjust the work plan and staffing so the engagement can be completed using approved methods. Meanwhile, your suggestion enters the evaluation process. If the proposed tool proves unsuitable, the team can explore alternatives that address the same need.
You have a clear way to contribute to improving the process. If a solution is approved and delivers measurable benefits, your contribution can be recognized through the firm’s innovation program, including an incentive where appropriate.
A strong technology policy provides a clear, practical path for innovation while maintaining compliance with professional standards and applicable legal and regulatory requirements. Firm leaders should support that balance through a timely, risk-based process for evaluating new tools, clear communication of expectations, and opportunities for employees to propose improvements. When professionals understand the reasons for safeguards and have a responsive process for pursuing new solutions, they have less incentive to bypass firm policies or use unapproved technology. Making responsible innovation accessible helps firms improve efficiency and service quality while upholding their professional obligations and preserving the trust clients place in their work.