AI Regulation Tracker  /  Professional standards

ICAEW Puts Binding AI and Technology Duties Into Its Code of Ethics

From July 1, 2026, a revised ICAEW Code of Ethics is in force. It adds technology to the list of things that can compromise objectivity, names automation bias as a professional-judgment bias, and spells out new threats from using technology in your work. This is a binding code for ICAEW members, not guidance, and it reaches cross-border firms and US-based members who hold the ACA.

By Anthony Guerriero  ·  The Leveraged Years Regulatory Research  ·  Published July 19, 2026  ·  Last updated July 19, 2026

On July 1, 2026, a revised edition of the ICAEW Code of Ethics came into force. The cover page says it plainly: this Code of Ethics applies from 1 July 2026, and early adoption was permitted before that date. Two things changed in this update. One is a broader rewrite of the role and mindset the Code expects of a professional accountant. The other, and the reason it belongs in an AI regulation tracker, is a set of provisions that put technology directly into the ethics rules that ICAEW members are bound to follow.

I want to be careful about what this is. The ICAEW Code is not a statute and it is not a government regulation. It is the conduct code of a professional body, and it binds the people who hold that body's qualification. For a chartered accountant with the ACA, that binding force is real. Breaching the Code is a disciplinary matter, not a debate. So when the Code now names automation bias or says objectivity can be compromised by over-reliance on technology, those are not talking points. They are duties.

Technology can now compromise your objectivity

Objectivity is one of the five fundamental principles, and the revised Code changed the wording of the objectivity rule itself. Under R112.1, an accountant must exercise professional or business judgment without being compromised by bias, conflict of interest, or, in the Code's own words, "undue influence of, or undue reliance on, individuals, organisations, technology or other factors."

Read that last clause slowly, because the word technology is new to it. The Code is telling members that leaning too hard on a tool, letting the software's answer stand in for your own judgment, is now framed as a threat to objectivity in the same breath as conflicts of interest. If you run a model, take its output, and sign off without the independent thinking the situation demands, you are not just being sloppy. You are potentially in breach of a fundamental principle.

A named list of technology threats

The Code works on a threats-and-safeguards model. You identify threats to the fundamental principles, evaluate them, and address them. The 2026 revision adds a new provision, 300.6 A2, that lists facts and circumstances relating to the use of technology that might create those threats. The examples it gives under self-interest threats are concrete:

• The data available might not be sufficient for the effective use of the technology.
• The technology might not be appropriate for the purpose for which it is to be used.
• The accountant might not have sufficient information and expertise, or access to an expert, to use and explain the technology and its appropriateness for the intended purpose.

That is a plain-language checklist for anyone deploying AI in accounting work. Is my data good enough. Is this the right tool for this job. Do I actually understand it well enough to use it and to explain it. The Code also flags a self-review angle: if the technology was designed or developed using the firm's own knowledge or judgment, that itself can create a threat. None of this bans AI. It requires you to think before you trust it, and to be able to show your reasoning.

Automation bias is now on the record

The Code has long acknowledged that bias affects professional judgment, and it keeps a list of biases members should watch for: anchoring, availability, confirmation, groupthink, overconfidence, and others. The 2026 update adds a new entry to that list at 120.12 A2. The Code defines it as "Automation bias, which is a tendency to favour output generated from automated systems, even when human reasoning or contradictory information raises questions as to whether such output is reliable or fit for purpose."

That definition is worth keeping on a wall. It captures the exact failure mode that AI creates in a finance workflow. The model produces a clean, confident answer, and the pull is to accept it even when something in front of you says it might be wrong. The Code now treats resisting that pull as part of exercising professional judgment properly.

IT services and the self-review line in assurance

The fourth piece sits in the assurance rules. When a firm provides an assurance engagement, taking on a management responsibility for the client's subject matter can create a self-review threat that independence rules are built to prevent. The revised Code, at 900.13 A4, gives worked examples of IT systems services that cross into management responsibility, such as hosting or being the only access point for the client's data and records, providing back-up or disaster-recovery for those records, or operating and monitoring the client's IT systems tied to the subject matter information.

Then 900.13 A5 draws the other side of the line. It clarifies that the ordinary collection, receipt, transmission, and retention of data provided by an assurance client, in the course of the engagement or to enable a permissible service, does not by itself amount to taking on a management responsibility. For assurance firms building AI and data pipelines into their engagements, that distinction is the difference between a normal service and an independence problem.

What this means for US accountants

The direct reach is the important part here. The ICAEW Code binds members of ICAEW, and plenty of them are in the United States. If you hold the ACA, or you work inside a global firm whose engagement teams include ICAEW members, this Code applies to that work regardless of where the desk sits. Cross-border and global-firm accountants do not get to treat this as a foreign rule that stops at the water's edge. It travels with the qualification. Applies to ICAEW members, including US-based ACA holders; it is a benchmark only for AICPA/US-CPA practice, which it does not bind.

If you are a US CPA with no ICAEW tie, this is not your rulebook. The AICPA Code of Professional Conduct governs you, and nothing about the ICAEW update changes a single US obligation. But it is one of the clearest signals yet of where professional ethics for accountants is going on AI. A major English-speaking body has decided that the way to handle AI is not a separate AI policy bolted on the side, but amendments to the core principles: objectivity, professional judgment, and the threats framework. That is a template. When the AICPA and the international ethics board work through their own AI provisions, expect the same moves, technology as an objectivity threat, automation bias as a named bias, and clearer lines on data and IT services in assurance. Reading the ICAEW text now is a way to see the shape of your own future duties early.

What to do now

If you or your firm hold the ACA, this is a compliance item, not a curiosity. Confirm who on your teams is an ICAEW member and treat the 2026 Code as live for their work. Build the 300.6 A2 questions into how you adopt any AI tool: data sufficiency, fitness for purpose, and whether someone can genuinely use and explain it. Name automation bias in your review procedures so that accepting a model's output without challenge is something a reviewer is expected to push on. And if you do assurance work, map your data-hosting and IT arrangements against the 900.13 A4 examples before an inspector does. For US-only CPAs, read the same provisions as a preview and start pressure-testing your AI governance against them, because the direction of travel is not subtle.

Browse the full AI Regulation News tracker

Informational analysis for working professionals, not legal, accounting, or audit advice. Confirm how any code or standard applies to your situation with qualified professionals in the relevant jurisdiction.