Certified public accountants have embraced artificial intelligence for routine drudgery like data entry, reclaiming hours for client advisory. Yet as ambitions grow toward analysis, reporting and marketing, pitfalls loom large. Ben Taylor, co-founder of SoftLedger, warns in Accounting Today that ‘AI comes with some risks – it can hallucinate and come to false conclusions when asked to extrapolate or reason through large amounts of information.’

Recent surveys underscore the rush: CPA.com’s 2025 AI in Accounting Report, released in June, reveals firms shifting from automation to augmentation, with Erik Asgeirsson, president and CEO of CPA.com, stating ‘AI is fundamentally reshaping the accounting profession, accelerating the move toward more strategic advisory services’ per CPA.com. But hallucinations persist, eroding trust in high-stakes finance where precision reigns.

Stanford and MIT researchers quantified gains in a Journal of Accountancy study, surveying 277 accountants and analyzing transactions at 79 firms. AI reallocates 8.5% of time from data entry to communication and assurance, enabling faster closes. Still, unchecked outputs risk audits and liability.

Hallucinations Undermine Core Reliability

Taylor’s first caution: Evaluate efficiency gains beyond basics. AI excels at drafting emails or QA reviews but falters in variable tasks like bank categorization, often demanding more rework. ‘Even if a model is 99.99% accurate, it can’t be a black box. In finance and accounting, the ‘why’ matters just as much as the ‘what,” Taylor writes. A Thomson Reuters survey cited in The CPA Journal shows only 1 in 10 professionals using generative AI, wary of errors in tax or audit work.

X user @SecretCFO echoes this, posting: ‘One problem: 10% of the time, they just make stuff up… In enterprise finance, where trust in the numbers is your oxygen, this isn’t a “quirk.” It’s a brick wall.’ UK accountants report 50% of firms seeing client losses from ChatGPT-style advice, per a survey highlighted by @MrEwanMorrison on X.

Forensic experts in CPA Practice Advisor note AI’s fraud detection promise but warn of oversight gaps, as the UK Financial Reporting Council found Big Four firms failing to monitor AI’s audit impact.

Audit Explainability: The Black Box Dilemma

Taylor’s second litmus: ‘Can I explain this to an auditor?’ Outputs must be documented; ‘If it isn’t documented, it isn’t done.’ Journal of Accountancy’s November piece positions CPAs as AI evaluators, with Supkis Cheek of Caseware saying CPAs’ rigor addresses AI’s trust deficit via SOC-like engagements.

AICPA & CIMA resources, including ‘Closing the AI trust gap,’ urge governance, as state lawmakers eye regulations per AICPA. Marta Zaniewski warns CPAs to engage, lest AI encroach on licensed services.

CPA Trendlines’ 2026 Outlook cites AICPA CEO Mark Koziel: ‘AI is not going to disrupt the accountant – it’ll change what the accountant does.’ Yet 72% of S&P 500 firms flag AI as a material risk for reputation and compliance, per Fortune analysis.

Delegation Demands: Scaling Without Overload

Third: Can juniors support AI outputs? Taylor advises pilots to delegate, scaling without CPA bottlenecks. Wolters Kluwer’s report shows AI adoption surging to 41% in 2025, with best firms holding quarterly oversight.

Texas CPA Magazine warns in ‘Assessing AI From a Tax Perspective’ that tools are ‘error-prone,’ demanding verification like staff work. @CynicalPublius, an attorney, notes on X: ‘AI ROUTINELY HALLUCINATES… Until the problem of AI hallucination can be eliminated, it is just so much puffery.’

Canadian accountants see tax errors piling from AI overreliance, per Yahoo Finance via @cagbaeze on X, mirroring U.S. trends.

Regulatory Pressures Mount

AICPA pushes AI literacy and assurance frameworks. CPA Canada’s Pamela Steer calls for human oversight: ‘Ensuring that humans remain central to critical decision-making mitigates risk.’ State bills preview federal moves, with CPAs pivotal in ethical AI.

Journal of Accountancy highlights agentic AI’s rise, predicting autonomous agents for audits by 2028, but urges CCPA/GDPR compliance.

World Economic Forum’s 2025 report flags accounting clerks’ decline due to AI, yet demands ‘AI and big data’ skills growth.

Cyber and Ethical Shadows

Confidentiality tops risks, per Accounting Today: AI tools expose data to breaches. CalCPA warns of biases perpetuating flawed decisions. CPA.com’s toolkit stresses security in strategies.

X’s @rryssf_ outlines liability: Disclose AI, face citation challenges; ignore, lose efficiency. GRF CPAs’ 2026 Risk Report: ‘The risk is not adopting AI, it is adopting it without a strategy.’

Firms must train, per Accounting Today, balancing rewards with WISP compliance to avoid PTIN revocation.

Strategic Path Forward

Taylor concludes: ‘Starting small, delegating and focusing on low-risk projects are all great ways to get more out of AI in 2026.’ CPA.com’s roadmap: Workflow automation with human verification. As @ShenHuang_ notes on X, vertical models like CPA-Qwen outperform generals on compliance.

CPAs lead by auditing AI, per Journal of Accountancy’s Carrie Kostelec. Amid Big Four disruption warnings from @MarioNawfal, those mastering guardrails thrive.

The profession evolves: AI augments judgment, but vigilance ensures integrity endures.