Investigation: UK Accountants Failing to Audit Their AI Tools

Summary of the Issue

The UK's largest accounting firms — including the Big Four (PwC, Deloitte, EY, KPMG) plus BDO and Forvis Mazars — are embedding AI tools into audits without formally measuring their impact on audit quality, as revealed by the Financial Reporting Council (FRC). Most firms track tool usage for licensing purposes, but not how well these tools perform in detecting risk or improving audit outcomes.

---

The Hidden Risks: From Business Fallout to Reputational Damage

1. Business Risk

• Faulty Audit Outcomes: Undetected errors caused by unverified AI tools could lead to regulatory penalties, investor lawsuits, or poor financial decisions.
• Algorithmic Bias or Failure: AI tools trained on biased or insufficient data can miss fraud or misstate risk, especially without human cross-checks.

2. Reputation Risk for Accounting Firms

• Repeat Audit Failures: Firms like BDO and Mazars have already been criticized for audit quality. Continued reliance on unchecked AI tools could lead to sanctions or client exodus.
• Public Perception: Firms claiming AI-powered efficiency without accountability could be seen as cutting corners, damaging trust.

3. Implications for HMRC and Tax Oversight

• If HMRC or similar bodies use AI to audit company filings or flag anomalies, errors in the AI’s judgment could wrongly trigger investigations or let bad actors slip through.
• Without standardised, auditable AI models, legal disputes around assessments may rise.

IT and computer science courses

---

Client Responsibility: Don’t Just Trust the Brand

If you're hiring an accountancy firm in 2025:

You need to ask:

• ❓ Are you using AI in your audit processes?
• 🤖 Which AI tools are being used — generative AI, machine learning, transaction scoring?
• 🛑 What controls or oversight do you have on these tools?
• 🛡️ What recourse do I have if the AI gets it wrong — and you miss something critical?

Why this matters: AI is not infallible, and if a firm uses automated systems without oversight, you as the client could still be held liable for compliance breaches, misstatements, or penalties. Always get clarity on who is accountable.

---

Could This Lead to a New Insurance Market?

Yes — and it may already be emerging.

AI Liability Insurance: The Next Big Niche?

• Insurers are beginning to explore “algorithmic risk” coverage — where businesses (including accounting firms) can insure themselves against AI-induced errors.
• Tech E&O (Errors and Omissions) policies exist, but AI-specific add-ons are still rare in accountancy.
• This could evolve into:
  • AI Audit Failure Insurance
  • Client Protection Addendums
  • AI Tool Certification Requirements (linked to insurance premiums)

Just as cyber insurance evolved alongside digital risk, AI insurance for financial processes is the logical next step — especially if FRC or HMRC begin to mandate formal oversight.

---

Solutions: What Should Be Done?

1. Internal AI Audits
   Firms must establish metrics and review protocols for every AI tool in use — especially where outputs influence audit outcomes.
2. Client Disclosures & Contracts
   Clearly define in engagement letters which tasks are handled by AI and who is responsible for validating results.
3. Third-Party Certification
   Independent firms or regulators could develop AI audit certifications, ensuring tools meet ethical, bias, and accuracy standards.
4. Regulatory Guidance & Insurance Linkage
   The FRC, ICAEW, and insurers should collaborate to create standards for insurable AI use, which could be tied to reduced premiums or mandatory certifications.

---

Final Thought

The idea that an audit can be performed by AI without anyone checking the checker is a recipe for disaster. For businesses and regulators alike, AI in accountancy must be treated like any other financial tool: with transparency, accountability, and oversight.

Hiring an audit firm? Don’t just ask for the partner’s CV — ask for the audit bot’s audit trail.

Latest AI News