Accounting | November 20, 2026

How AI Risks Can Affect Accounting Firms’ E&O Insurance

Commercial insurers increasingly factor in how businesses use AI when setting premiums and coverage terms. For US accounting firms, securing lower premiums depends on how effectively they manage AI-related risks and demonstrate those controls to insurers. AI risks are becoming an important consideration in commercial insurance underwriting. 2026 research by ScienceSoft, an AI transformation company... Read more »

Vital Soupel

Commercial insurers increasingly factor in how businesses use AI when setting premiums and coverage terms. For US accounting firms, securing lower premiums depends on how effectively they manage AI-related risks and demonstrate those controls to insurers.

AI risks are becoming an important consideration in commercial insurance underwriting. 2026 research by ScienceSoft, an AI transformation company working with the insurance industry, predicts that 60–80% of new policies and renewals across errors and omissions (E&O) and cyber insurance lines will incorporate AI-related risks into underwriting by 2028.

Accounting firms have good reason to pay attention: the latest domain studies indicate that roughly 87% of accounting service providers already use at least one AI tool. With E&O insurance maintained by the vast majority of top-tier and midsize US certified public accountant (CPA) firm, and 88% of CPAs reporting having dedicated cyber liability coverage, a major share of the industry is likely to face AI-specific underwriting scrutiny.

Yet introducing AI into accounting services doesn’t automatically expose your firm to higher insurance premiums. What matters is how well you govern AI use, control the associated risks, and demonstrate those safeguards to insurers. According to ScienceSoft’s research, differences in these practices can result in a 15–40% variation in premiums for AI adopters.

The following four measures can help your firm strengthen its AI risk profile and make a more credible case for favorable premium terms.

1. Build an AI inventory and establish clear AI usage boundaries

An AI inventory gives underwriters a clear picture of your firm’s AI use and helps them distinguish between AI applications with limited exposure and those that could lead to costly E&O or cyber claims. This allows for a more accurate assessment of your firm’s risk profile.

Your inventory should cover all AI capabilities introduced across the organization, including those embedded in existing software. Classify AI use cases by risk, considering the sensitivity of the data they process, their level of autonomy, and the potential impact of errors. Drafting internal meeting agendas without confidential data using an AI assistant would receive a low-risk score, while using largely autonomous AI agents for tax research would receive a higher risk score, with other applications falling somewhere in between.

For each AI solution, record its provider, purpose, and data access permissions. Also document whether AI can act autonomously and who is responsible for overseeing it. Then establish a written policy specifying which tools are approved, what data employees may enter, and which activities require supervision or are prohibited. Beyond providing evidence for underwriting, this also gives accountants clear guidance on appropriate AI use and helps reduce the risk of misuse stemming from limited AI literacy.

This approach aligns with generative AI risk management guidance published by CAMICO, a major US insurer specializing in CPA firms. The guidance recommends documenting authorized AI use, conducting provider due diligence, and training employees on responsible AI practices.

2. Make human review mandatory for client-facing AI outputs

Even highly capable AI can occasionally produce incorrect results. Without dedicated quality control checks, those errors may reach clients and expose your firm to professional negligence claims. This is particularly relevant for agentic AI systems that can autonomously pass outputs between service delivery stages and potentially interact directly with clients. ScienceSoft’s research suggests that AI risk underwriting will become more granular as AI systems gain autonomy.

CAMICO recommends verification by qualified professionals as an important safeguard against AI-related errors. The AICPA Member Insurance Program likewise advises supervising AI output as you would a team member’s work. These checks may limit some efficiency gains from AI automation, but they are a worthwhile trade-off for maintaining accuracy in client-facing work and containing the risk of E&O losses.

Automated, rule-based validation can serve as the first line of AI control. Such services can check AI outputs and calculations against predefined rules and flag gaps for human review. This lets accountants focus on exceptions and decisions requiring professional judgment rather than checking every routine output. At the same time, AI-generated tax returns and financial advice should still undergo mandatory review by qualified accountants, while complex tax positions or valuation assumptions should ideally receive senior-level approval.

Document your AI output review process, too. Preserve workflow logs, corrections, and approvals within engagement files. These records give underwriters tangible evidence of your firm’s AI oversight and help demonstrate the care your firm exercised if a claim arises.

3. Secure sensitive data before introducing it into AI workflows

Data confidentiality is one of the biggest AI-related cyber insurance concerns for accounting practices. Your employees routinely handle Social Security numbers, tax returns and bank account details. Entering this information into external AI services without appropriate safeguards can create risks related to data storage, processing, and disclosure, potentially worsening your firm’s insurance risk rating.

Your data security setup should include mechanisms for data encryption, multi-factor user authentication, and role-based access controls for both accounting employees and AI tools. Where possible, minimize and anonymize sensitive data before submitting it to AI applications; AI firewall tools available on the market can help automate this process. Also, consider training employees to recognize AI-related data privacy risks and avoid inadvertently disclosing confidential information through AI tools. Documenting this training can provide additional evidence of your firm’s security controls during underwriting.

Before approving an external AI service, evaluate where data is stored, whether it is used for model training, and how long it is retained. Review who can access your data and how security incidents will be reported. Legal requirements make vendor due diligence particularly important: for tax services, Internal Revenue Code Section 7216 and its regulations restrict the use and disclosure of tax return data by tax return preparers. Sharing this data with third-party AI providers may therefore raise compliance concerns depending on applicable exceptions and consent requirements. CAMICO also identifies Section 7216 as an important consideration for CPAs evaluating third-party AI tools.

Documented vendor risk assessments, security reviews, and contractual data protection commitments will give underwriters evidence of your due diligence and compliance efforts.

4. Monitor AI performance and document corrective actions

An AI tool that initially performs well may produce less reliable results over time as its operating conditions change or its underlying models are updated. Ongoing monitoring helps your firm identify AI performance drifts before they lead to operational errors or security incidents. You can establish this oversight through your existing enterprise monitoring platforms or dedicated AI observability tools like Amazon CloudWatch or LangSmith; there are plenty of options available on the market.

Set dedicated performance and risk KPIs for each AI-supported workflow and track changes over time. For document intelligence tools, monitor trends in processing error rates and the frequency of reviewer corrections. For AI-assisted reconciliations, track changes in transaction-matching accuracy and exception rates. AI agents that continuously process incoming invoices or post routine accounting entries may also warrant real-time alerts when error rates spike or unusual activity occurs so that employees can intervene promptly.

You should also establish a response procedure for AI-related incidents. For instance, if an AI tool repeatedly misclassifies transactions, a responsible manager should suspend its use, identify affected engagements, and override erroneous entries. This helps prevent isolated errors from spreading across multiple client records or reporting periods.

Keep records of monitoring results, incidents, and corrective actions. Here again, you can use monitoring and observability tools to automatically maintain traceable AI activity and performance logs. These records show underwriters that your firm doesn’t simply put AI safeguards in place but regularly verifies that they remain effective.

AI risk management affects more than just insurance prices

A strong AI risk mitigation program with well-documented oversight won’t automatically qualify your firm for a premium discount. Pricing will still depend on factors like your service scope and claims history, as well as coverage limits and insurer appetite.

However, effective AI controls and supporting evidence may help you challenge an unfavorable risk assessment, negotiate better coverage conditions, and keep renewal costs under control, particularly when your other risk factors remain unchanged.

Ultimately, these measures aren’t just about insurance costs. They also help you keep AI-related risks under control and safely make the most of AI-powered accounting automation.

===

Vital Soupel is a Senior Insurance IT & AI Consultant at ScienceSoft, an AI transformation and software development company specializing in the insurance industry. His work spans insurance AI strategy and program planning, AI solution design and implementation, and operational transformation consulting. Website: https://www.scnsoft.com/insurance.

Sign in to get access to this free resource, and all of our whitepapers and reports.

Download this content today!

Register to get free access to this content, as well as newsletters, continuing education, podcasts, and more…

Leave a Reply