Can CAs Offer AI Consultancy? What the 2026 Code of Ethics Allows
Yes. ICAI's 2026 Code of Ethics has expanded Management Consultancy and Other Services to include Artificial Intelligence (AI) Consultancy in areas of services which can be rendered by a Chartered Accountant in practice.
That is a small line with a large commercial implication. CA firms can now discuss AI advisory more openly as part of their professional service mix, provided the work stays within what a CA is competent and permitted to render, and provided audit-independence restrictions are respected.
Facts checked: 26 August 2026 against ICAI's revised Code announcement and the August 2026 "Know Your Ethics" page in The Chartered Accountant Journal.
What changed in the Code of Ethics 2026
ICAI announced the revised 13th edition of the Code of Ethics on 30 May 2026. Volumes I, II and III apply from 1 April 2026, except the social-impact and sustainability-assessment item that applies from 11 December 2025.
The August 2026 "Know Your Ethics" note in The Chartered Accountant Journal adds an important practical clarification. It says the Code has expanded the scope of Management Consultancy and Other Services under Section 2(2)(iv) of the Chartered Accountants Act, 1949 to include:
- forensic accounting and investigation,
- research analyst services recognized by a regulator,
- social impact, CSR impact and BRSR-type assessment and evaluation,
- and Artificial Intelligence (AI) Consultancy in areas of services which can be rendered by a Chartered Accountant in practice.
For firms building advisory lines around CA GPT, AICA, Tally automation, audit analytics or AI governance, that last phrase matters.
What AI consultancy can reasonably mean for a CA firm
The safest reading is not "CAs can do any AI project." It is narrower: CAs can provide AI consultancy where the underlying professional domain is one CAs can serve.
Examples that fit naturally:
| AI consultancy service | Why it fits a CA firm |
|---|---|
| AI readiness assessment for finance and audit teams | Governance, process and control review |
| AI policy for client data and DPDPA boundaries | Data protection, confidentiality, professional conduct |
| Accounting workflow automation scoping | Books, reconciliations, GST, TDS, MIS |
| Tally / ERP data-quality diagnostics | Accounting-system and audit-readiness work |
| Audit analytics implementation support | Audit methodology and evidence workflow |
| AI vendor evaluation for finance functions | Control, data, assurance and ROI assessment |
| Staff training on responsible AI use in audit/tax | Professional education and quality management |
Examples that need caution:
| Service | Caution |
|---|---|
| Building a production AI app for a client | Needs technical competence, contracts, support model and cyber risk controls |
| Implementing AI in a statutory audit client | Independence and self-review risk must be assessed |
| Certifying an AI model as "accurate" | CAs need a defensible assurance framework before making such claims |
| Uploading client data into public AI tools | DPDPA, confidentiality and client-permission issues arise immediately |
The independence boundary
AI consultancy becomes sensitive when the client is also an audit client.
The core question is whether the firm is designing, implementing or operating a system that it will later audit. If yes, the self-review risk can become serious. A firm can advise on policy, training, data discipline and selection criteria more safely than it can build and operate the client's finance AI system and then audit its outputs.
A practical test:
- Are we making a management decision?
- Are we designing a control that we will later test?
- Are we preparing accounting records or source schedules?
- Are we implementing a system that affects financial reporting?
- Would an informed third party think we are reviewing our own work?
If the answer is yes, the firm needs safeguards, scope limits, separate teams, or refusal.
How to package AI advisory without overclaiming
The cleanest first offering is not "AI transformation." That is too broad and usually under-scoped.
Better packages:
1. AI use policy for CA firms and finance teams
Deliverables:
- approved / restricted / prohibited AI tools list,
- client-data redaction rules,
- prompt and output retention policy,
- reviewer sign-off requirements,
- DPDPA and confidentiality checklist.
2. Finance process automation assessment
Deliverables:
- process map,
- automation opportunities,
- data-readiness score,
- tool/vendor shortlist,
- implementation roadmap,
- control and audit-trail requirements.
3. Audit analytics pilot
Deliverables:
- one ledger or GST/TDS dataset,
- defined procedures,
- exception logic,
- auditor review notes,
- evidence pack,
- limitations statement.
4. AI training for audit and tax teams
Deliverables:
- prompt discipline,
- hallucination checks,
- source verification,
- client-data boundaries,
- working-paper documentation.
These are concrete, reviewable and aligned with a CA firm's professional strengths.
How AICA Level 2 fits
AICA Level 2 now makes more sense in this context. Its current Ver 3.0 syllabus includes Python, app deployment, Tally-Claude MCP, local models, n8n automation and agentic workflows. That is not a general manager course; it is a builder course for CAs who may actually deliver AI consultancy.
So the practical path is:
- Level 1 for baseline literacy.
- ICAI AI publications for policy and professional context.
- Level 2 only for the person who will build or advise technically.
- Tool/vendor training for execution.
- Internal independence and quality review before client rollout.
See the AICA certification review and the ICAI AI publications guide.
Website and marketing implications
The Code of Ethics 2026 also modernizes how CAs present themselves online. ICAI's August 2026 ethics note says members can mention the firm's name and contact details in educational podcasts or videos, conduct webinars, and in certain cases mention client names in write-ups with permission.
That creates a legitimate content opportunity for CA firms:
- publish educational AI explainers,
- run client webinars,
- show firm capabilities,
- mention peer review / AQMM status where permitted,
- and explain AI advisory services without making exaggerated claims.
The marketing line should stay factual: "AI policy and audit-analytics advisory for finance teams" is safer than "we guarantee AI transformation."
Frequently Asked Questions
Can a CA firm offer AI consultancy after the Code of Ethics 2026?
Yes, the August 2026 ICAI ethics clarification says AI Consultancy has been added under Management Consultancy and Other Services, in areas of services which can be rendered by a Chartered Accountant in practice.
Can a statutory auditor implement AI systems for an audit client?
Only with extreme care. If the system affects financial reporting or controls that the firm will later audit, self-review and independence threats arise. The firm should assess scope, safeguards and whether the service should be refused.
Can CAs advertise AI consultancy on their websites?
The revised Code has modernized website and write-up rules, but claims must remain factual, not misleading, and within the Code's advertising and professional conduct boundaries.
Is AICA Level 2 required to offer AI consultancy?
No, it is not a legal prerequisite. But for a firm that intends to build or deploy AI workflows, AICA Level 2 or equivalent technical training is a practical competence signal.
Sources: ICAI announcement on Revised Code of Ethics, 13th edition, ICAI Chartered Accountant Journal - Know Your Ethics, August 2026, ICAI Disciplinary Directorate Code of Ethics page.
Try CORAA -> Use audit-grade AI without turning every CA into a software engineer: Tally import, full-population scrutiny, GST/TDS checks and working papers with review trails. See pricing · AI Lab · AI tool evaluation checklist.