AI in Corporate Governance: 2026 Board Playbook

Last verified: 2026-07-17

AI in corporate governance is not a boardroom gadget or a line item in the technology report. It is a working method a director applies to the actual job: reading a dense board pack the night before, testing whether management’s strategy survives a downturn, catching the risk buried in an appendix, and asking the question that changes the discussion. The same assistant that helps one director walk into the audit committee sharper will, for the director at the next company, produce a confident summary with a number that was never in the accounts. The difference sits almost entirely in how the tool is used and how clearly the director’s own accountability is drawn, not in which vendor supplied it.

Consider two non-executive directors on similar boards, both handed the same enterprise AI assistant this year. The first treats it as an oracle: pastes a draft risk report into a public chatbot, asks “what should the audit committee worry about,” and forwards the answer with light edits. The output reads well and flags a control gap the model invented, which unravels the moment the head of internal audit is asked about it in the meeting. The second treats it as a fast, tireless analyst that is never the final word: feeds it the anonymised pack, asks it to list the three questions the papers do not answer, drafts a challenge to the growth assumptions, then checks every figure against the source before saying a word. Same tool, same board cycle, opposite result.

That gap is the whole subject of this guide. Most directors were handed a chatbot and a vague encouragement to “understand AI,” with no model for what a genuinely useful governance application looks like or where the guardrails belong. They ask it to summarise a report, get a serviceable summary, and quietly conclude AI is overhyped. Meanwhile the directors pulling real value from it are doing something narrower and more deliberate, and they are just as clear about the second job the technology creates for them: overseeing the company’s own AI on behalf of shareholders.

The demand side is already moving. According to Deloitte’s 2025 Global Boardroom Program survey of 700 board members and executives across 56 countries, 31% of respondents say AI is still not on the board agenda, down from 45% a year earlier, while 66% admit their boards have “limited to no knowledge or experience” with it. In India the workforce backdrop is sharper still: the Microsoft and LinkedIn 2024 Work Trend Index found 92% of Indian knowledge workers already use AI at work. So the question for a director in 2026 is not whether the technology matters. It is whether you will use it to govern better and oversee it competently, or leave it to management and hope.

This guide sets out exactly how, step by step, with copy-ready prompts for a director’s work and the guardrails that keep the whole thing inside your duties.

Board directors use AI in corporate governance two ways: to sharpen their own oversight (interrogating the board pack, benchmarking peers, pressure-testing strategy and risk, and asking better questions) and to govern the company’s own AI (setting policy, risk appetite, and the oversight questions management must answer). Both rest on one rule: AI drafts and challenges, but the director verifies every fact and keeps the statutory duty that no model can hold.

One boundary before we start. This guide is about a director using AI in governance work and overseeing AI competently, not about getting the whole organisation to adopt it, which is a separate change-management job covered in the companion piece on how senior leaders drive AI adoption across teams. For the day-to-day decision and productivity uses that overlap with executive work, AI for managers is a useful companion, and if you want to go deeper on the prompt craft itself, the guide to prompt engineering for executives is the natural next read. Here, the focus is the boardroom.

The two jobs AI creates for a board director

AI in corporate governance lands on a director’s desk as two separate jobs, and confusing them is the first mistake. The first is using AI as a tool in your own oversight work. The second is overseeing the company’s use of AI as a matter the board is accountable for. They call for different skills and different guardrails, and a director who runs them together tends to do neither well. Name them apart and each becomes manageable.

Job one: the four oversight tasks AI actually helps with

In your own work, AI helps with four specific tasks, and deciding which one you are doing before you open the tool is what keeps it useful. It digests the board pack, turning two hundred pages into a structured brief of what changed, what is missing, and what deserves a question. It researches and benchmarks, pulling peer governance practice, a competitor’s disclosure, or the shape of a new regulation into a form you can scan. It pressure-tests strategy and risk, arguing against a plan or running a pre-mortem on a proposed acquisition. And it sharpens the questions you bring to the table, which is the part of oversight that actually moves a company.

Each of these plays to a language model’s strength: reading, structuring, and challenging text, at speed, without tiring. None of them asks the model to be right about your business on its own, which is exactly why they are safe when paired with your judgment. The Harvard Law School Forum on Corporate Governance describes much the same set: summarising board materials and surfacing insights, benchmarking peer practice, and scenario modelling to test strategy.

Job two: overseeing the company’s own AI

The second job is heavier, because here the board is not using a tool, it is answering for one. When a company deploys AI in lending, hiring, pricing, or customer service, the board carries the oversight of the risks that creates, and that accountability does not transfer to the vendor or the model. A director cannot point at an algorithm when a biased decision or a data breach reaches the regulator. The duty to oversee stays with the humans on the board.

This is why the two jobs must not blur. Getting fluent with AI as a personal tool is useful, but it is not the same as governing the company’s AI, and a board that has done the first sometimes assumes it has done the second. It has not. The oversight questions, the policy, and the risk appetite in the second half of this guide are the real deliverable, and they are what an investor, an auditor, or a regulator will judge the board on.

What the data says boards are actually doing

The picture is one of fast recognition and slow capability. Deloitte’s 2025 survey found 40% of directors say AI has made them think differently about their board’s composition, a sign the recognition has arrived. Yet PwC’s 2025 Annual Corporate Directors Survey found only 35% of directors say their boards have brought AI into their oversight role, so most have not yet turned recognition into practice.

The expertise gap is starker. MSCI Institute analysis found that as of mid-2025, 25% of large and mid-cap boards reported at least one director with AI expertise, up around ten points in three years, but only 14% had put that expertise to effective use, and roughly 2% of individual directors are identified as AI experts. Read together, the numbers say the same thing: the board that builds real competence here, rather than appointing one specialist and moving on, is still doing something most of its peers have not.

Set up AI before you bring it into the boardroom

Before you put AI to work in governance, spend an hour on a setup that saves you from the two errors that sink most early attempts: leaking confidential board information, and getting thin output from lazy prompts. Most directors skip this and type questions into a public chatbot, which is why the first attempt underwhelms and the company secretary gets nervous. The setup is three decisions: which tools, what never goes in, and how to write a prompt that earns its keep.

Advertisement

Pick approved tools and draw the boardroom data line

Start with the tools, then draw the data line, because a board handles some of the most sensitive information in the company. For most directors, one general assistant covers the drafting and analysis work, ideally the enterprise version your organisation or your board portal provider has approved, because those carry contractual data protections the free consumer tiers do not. If the company has sanctioned a specific tool or a governance-grade feature inside the board portal, use that for anything touching company information.

The data line matters more than the tool choice, and in the boardroom it is strict. Board papers, material non-public information, unpublished results, deal terms, privileged legal advice, and anything covered by insider-trading rules never go into a tool that is not on the company’s approved, contractually covered list. Pasting a draft results statement or an acquisition memo into a public chatbot is not a shortcut, it is a potential disclosure breach and, in a listed company, a securities-law problem with your name on the attendance sheet.

So what is safe to feed it? Anonymised figures with entity names replaced by placeholders, structural and methodology questions, published filings, public governance codes, and your own rough notes with the identifying detail stripped. Swap the real company for “Company A,” mask the sensitive numbers, and you keep almost all the analytical value with almost none of the exposure.

Write a director-grade prompt: role, context, task, format

The single skill that lifts your results is prompting, and a usable director prompt has four parts: role, context, task, and format. Most weak output traces back to a one-line question with no context, which is exactly what a busy director tends to type between meetings. Give the model a role to play, the background it needs, the specific task, and the shape you want the answer in, and the quality jumps.

Here is what that looks like for a real governance task, interrogating a board pack:

You are an experienced non-executive director and audit committee chair. Context: I have the quarterly board pack for a mid-market manufacturing company. Revenue is 6% below plan, the pack blames “market conditions,” and net debt has risen against a covenant I care about. Task: based on the anonymised summary I paste below, give me the eight sharpest questions I should ask management, ranked by importance, and for the top three explain what a weak answer would sound like. Format: a numbered list, each question in one sentence, no preamble. Here is the summary: [paste anonymised notes].

Notice the difference. The role sets the register, the context supplies the real tension in the numbers, the task is specific and bounded, and the format makes the output usable in the meeting without a rewrite. If you want to build this skill properly, the guide to prompt engineering for executives goes deeper on the patterns that hold up under pressure.

Draw the accountability line before you lean on the tool

Decide now what a director’s judgment never delegates, because deciding it in the moment is how a duty quietly erodes. Some things belong to an accountable human by law. Under Section 166 of the Companies Act, 2013, a director must exercise their duties with due and reasonable care, skill and diligence, and must exercise independent judgment. A model can inform that judgment. It cannot hold it, and no board minute should ever read as though it did.

So the line for your role is simple to write and worth writing down: AI can summarise, research, challenge, and draft, but the director reads the source, forms the view, and owns the vote. Approval of accounts, the assessment of a going-concern question, the decision to back or block a transaction, and any conclusion the board records as its own stay with the person, not the software. Set that line in advance and you can use the tool freely everywhere else.

Use AI to sharpen your board oversight

Using AI to oversee better is the highest-value job in your own toolkit, and it is where most directors underuse it. So why do so few use it this way? Habit, mostly: the chatbot arrived as a writing aid, and “analytical sparring partner for a board pack” never occurred to them. Used well, a model reads faster than you can, never gets bored on page 180, and has no loyalty to management’s framing. It does not form the board’s view. It makes the view you form better informed.

Interrogate the board pack and surface the missing questions

The board pack is where AI pays off first, because the volume is the problem and reading is the model’s strength. A good director already knows the job is not to absorb every page, it is to find the two or three things that matter and the questions management would rather not field. AI compresses the reading so you spend your time on the judgment, which is the part that cannot be delegated.

Beyond the interrogation prompt above, ask the model to compare this quarter’s narrative against last quarter’s, using the anonymised summaries: what has management stopped mentioning, which risk has quietly moved from “emerging” to nowhere, where has the language softened around a number. That drift is often where the real story sits, and it is exactly the kind of pattern a tireless reader catches and a rushed one misses. You still verify anything it flags against the actual pack before you raise it.

Pressure-test strategy and risk with a pre-mortem

A pre-mortem is the single most useful strategy prompt a director can run, and AI makes it effortless. The technique is old: before the board backs a plan, imagine it has already failed and work backwards to why. AI suits it because it has no stake in management’s proposal and no reluctance to contradict the executive who authored it. Feed it the plan, anonymised, and let it attack.

Here is a prompt you can adapt:

You are a skeptical audit committee chair and a former CEO. Here is the strategy management is asking the board to approve: [paste the plan with confidential detail masked]. Assume it is now two years later and the plan has clearly failed. Give me the eight most likely reasons it failed, ranked by probability, and for the top three, tell me the early warning indicator the board could watch each quarter to catch it before it is too late.

What comes back is a risk map you can bring to the discussion before the board commits, not a post-mortem after the write-down. Some of the eight will be generic, so keep the three or four that are genuinely about this business and this plan. For a broader view of AI in structured decision-making, the companion on AI for managers covers the decision-support patterns in more depth.

Benchmark governance practice against peers and the code

AI is fast at turning “how do others handle this” into a structured comparison you can actually use in a nominations or governance committee. Directors often carry a vague sense that a peer does board evaluation or ESG disclosure differently, without the time to pin it down. The model builds that comparison from public sources in minutes, and you verify the specifics against the original filings.

Try this:

I chair the nomination and remuneration committee of a listed Indian company. Using publicly available governance disclosures, build a comparison table of how three comparable listed companies structure their board evaluation process: who runs it, how often, whether it is externally facilitated, and how findings are disclosed. Flag any practice that looks stronger than a standard annual internal review, and cite the public source for each so I can verify it.

The table is a starting point for the committee, not a conclusion. Its value is that it turns a hunch into a specific, sourced comparison you can test, and the sourcing requirement is what lets you check the model has not invented a practice that does not exist.

Beware the confident, discoverable, wrong answer

The real danger in AI-assisted oversight is not a single wrong output, it is trusting a fluent machine too readily, and doing so in a record that can later be read against you. A model produces well-structured, authoritative text whether it is right or inventing, and the polish is precisely what disarms the scrutiny a careful director would normally apply. Harvard’s corporate governance analysis names both risks directly: hallucinated outputs that look credible, and the written record a director creates by typing exploratory questions into a tool.

So verify anything with a fact or a figure in it before you act on it or repeat it, and treat every AI prompt and answer as potentially discoverable. Tie figures back to the source pack, check any legal or regulatory claim against the actual rule, and keep your queries to secure, approved environments where the record is governed by a retention policy. The model is a strong first reader and a weak final authority, so keep it in the first chair, never the last.

Oversee the company’s AI from the boardroom

Overseeing the company’s own AI is the job the board cannot delegate and the one regulators are starting to ask about directly. Where the first half of this guide made you a better user of AI, this half is about your duty as a steward of the company’s use of it. A board does not need to understand the mathematics of a model. It does need to know who is accountable, what could go wrong, and whether management can answer for it.

The oversight questions every board should put to management

Effective oversight of AI starts with a short list of questions the board asks management and expects clear answers to. Who owns each significant AI system and its outcomes? What data does it use, and does that use comply with data-protection law? Can the company explain a material decision the system made, to a customer or a regulator? And where is the human in the loop for decisions that affect people’s money, employment, or rights?

If management cannot answer these cleanly, that is itself the finding, and it belongs in the minutes. The board’s role is not to audit the code, it is to satisfy itself that a competent, accountable process sits behind every AI system that carries real risk. A director who keeps returning to these four questions, meeting after meeting, is doing the oversight job whether or not they can read a line of Python.

Build board-level AI governance: policy, risk appetite, reporting

Ad-hoc oversight does not survive contact with a real incident, so the board’s second task is to insist on a governance framework it can point to. That means a board-approved AI policy, a stated risk appetite for where and how AI may be used, a clear line of management accountability, and a regular reporting cadence so AI risk reaches the board before it reaches the front page. India’s securities regulator is already moving this way: SEBI’s proposed framework for AI use by regulated entities makes the entity solely responsible and liable for the outputs of its AI and for the integrity of investor data, and its proposals point toward a board-approved governance framework, model explainability, and human oversight of material outputs.

The framework does not need to be long, it needs to be owned. A board that has approved a specific policy, set a risk appetite, and receives real reporting on AI has converted a vague anxiety into a governed process. Getting the wider organisation to actually work within that policy is its own change-management effort, which the companion guide on driving AI adoption across teams addresses.

Know the rules that bind the board

A director overseeing AI needs a working grasp of the rules that now attach to it, because ignorance of them is not a defence the board wants to offer. In India the Digital Personal Data Protection Act, 2023 governs how personal data may be used to build and run AI systems, and its penalties for breaches reach the company the board oversees. SEBI’s expectations bind regulated entities, sectoral regulators such as the Reserve Bank of India add their own, and any company that offers AI-driven products or services in the European Union, or whose AI outputs are used there, can fall within the reach of the EU AI Act regardless of where it is based.

None of this requires the board to become a compliance function. It requires directors to know enough to ask whether management has mapped the applicable rules and can show the company meets them. The underlying duty is not new, it is the same duty of care, skill and diligence that already sits behind every other risk the board oversees, and the stakes of getting it wrong are close to those explored in this analysis of independent director legal liability in India. AI simply adds a fast-moving, high-consequence risk to the list the board already owns.

Mistakes board directors make with AI

Most of the damage from AI in the boardroom comes from a short list of predictable mistakes, and knowing them upfront is far cheaper than learning them in a regulatory inquiry. So which ones actually cost you? Four, mostly. The tool is forgiving on low-stakes drafting and unforgiving on facts, confidential information, and the board’s own record, which is precisely where governance lives. All four are avoidable.

Trusting a fluent summary as verified fact

The trap that catches experienced directors is mistaking fluency for accuracy. AI writes with total confidence whether it is right or inventing, and the polish is exactly what disarms the scrutiny a seasoned board member would normally apply. It will state a covenant threshold, a peer’s practice, or a regulatory requirement that is subtly or completely wrong, in the same assured tone as the correct material, and a director who repeats it in the meeting owns the error.

So verify anything with a fact or figure before it leaves your notes. Tie it back to the pack, the filing, or the rule, and read AI output as a claim to test, not an answer to trust. This is the step busy directors skip first under time pressure, and it is the one that turns a helpful brief into a misstatement on the record.

Feeding board papers or MNPI into public tools

The most serious error is putting confidential board information into a tool that is not cleared for it. It is easy to do under pressure the night before a meeting: you paste a draft results statement or a deal memo to get a faster read, and now that information has left the company’s control and possibly breached insider-trading or disclosure rules. “I did not realise the tool was not approved” is not an explanation any director wants to give a regulator or a fellow board member.

The fix is the data line from earlier, applied without exception. Strip or mask the identifying detail, use placeholders, or use the approved enterprise tool for anything sensitive. It costs a few seconds and prevents the kind of incident that becomes a board matter with your name attached.

Creating a discoverable written record of half-formed queries

The subtle mistake, and the one lawyers warn about, is forgetting that your prompts and the model’s answers can become a record. A director exploring a worry by typing speculative questions into an AI tool may be creating written material that an adversary in litigation or an investigation could later read in the least charitable light. An exploratory “could this be fraud” prompt is not the same as a considered board discussion, but it can look damaging out of context.

Keep exploratory queries inside secure, company-approved environments governed by a retention policy, and keep the board’s actual reasoning where it belongs, in properly minuted discussion. The point is not to stop thinking with the tool, it is to be as deliberate about the record you create with AI as you already are with email.

Drifting from oversight into management

The mistake unique to this topic is letting AI-fuelled detail pull the board across the line from oversight into management. When a model hands a director a granular analysis of an operational issue, the temptation is to take it into the meeting and start running the function, which is management’s job, not the board’s. Harvard’s analysis flags this overreach as a distinct risk of AI in the boardroom.

The discipline is to use the extra insight to ask a better question, not to seize the executive’s decision. If AI helps you see that a division’s margins do not add up, the board move is to press management for an explanation and a plan, not to redesign the pricing yourself. Sharper oversight is the goal; becoming a shadow management team is the failure mode. The table below sums up the split that keeps you on the right side of it.

A 90-day plan to bring AI into your governance work

You can go from occasional dabbler to genuinely competent across one board cycle, and here is the plan to do it. Reading about AI in corporate governance changes nothing; using it on real governance work for a quarter changes how you prepare, what you ask, and how the board oversees the company’s own AI. The plan is deliberately staged, because a director’s calendar runs in quarters and one capability at a time is what actually sticks.

One board cycle, staged

The ninety days break into three moves, each building on the last. In the first month, set up: pick the company-approved tool, write your boardroom data line and your accountability line against your duties, and get the company secretary comfortable with how you will use it. In the second month, apply it to your own oversight: run the board-pack interrogation and one pre-mortem on a real strategy item, editing and verifying everything before you raise it. In the third month, turn to the second job: put the four oversight questions to management, and press for a board-approved AI policy, a stated risk appetite, and a reporting cadence if the company does not have them.

Why stagger it? Because trying all of it before one board meeting is how directors abandon the effort by the next. One capability per month, applied to work you were doing anyway, compounds into competence without adding load. By the end of the cycle you will have used AI across your own oversight and started the board on governing the company’s AI, with your duties intact from day one rather than reconstructed after an incident.

Measure whether it is working

Track outcomes, not activity, or you will mistake a busy quarter for a better-governed one. Logging into the tool between meetings proves nothing. The honest questions are whether your oversight got sharper and the board’s grip on AI risk got firmer: did you walk into the audit committee with better questions, did a pre-mortem surface a risk the pack had buried, did management’s answers to the four oversight questions actually improve.

Pick two or three concrete markers and watch them across a couple of cycles. The quality of the questions the board asks. Whether AI risk now reaches the board on a schedule rather than after an event. Whether the board evaluation shows members feel better informed. If the markers move, extend what you route through AI and deepen the oversight. If they do not, tighten your prompts and your verification rather than blaming the tool, because the gap is almost always in the instruction or the inputs, not the model. Deloitte’s data is a useful reminder: recognition is common now, but capability still separates the boards that govern AI from the ones that only worry about it.

Frequently asked questions

How do board directors use AI in corporate governance? Two ways. In their own oversight work, they use AI to digest the board pack, benchmark peer governance practice, pressure-test strategy and risk with pre-mortems, and sharpen the questions they ask management. As stewards, they oversee the company’s own AI by setting policy, defining a risk appetite, and holding management to clear answers on accountability, data, explainability, and human oversight. In both cases the model drafts and challenges, and a director verifies every fact and keeps the statutory duty.

Can AI replace the judgment of a board director? No. Under Section 166 of the Companies Act, 2013, a director must exercise independent judgment and act with due care, skill and diligence, and that duty cannot be delegated to a model. AI can inform the judgment by summarising, researching, and challenging, but the director reads the source, forms the view, and owns the vote. A board decision recorded as the board’s own must be exactly that.

Is it safe to put board papers into an AI tool? Only within strict limits. Board papers, material non-public information, unpublished results, deal terms, and privileged advice should never go into a tool that is not on the company’s approved, contractually covered list, because doing so can breach insider-trading, disclosure, or confidentiality obligations. Anonymised figures, public filings, governance codes, and masked notes are generally safe. When in doubt, strip the identifying detail or use the enterprise tool the company has cleared for sensitive work.

What is the board’s role in overseeing the company’s AI? To satisfy itself that a competent, accountable process sits behind every AI system carrying real risk, not to audit the code. That means insisting on a board-approved AI policy, a stated risk appetite, clear management accountability, and regular reporting so AI risk reaches the board early. The practical test is whether management can answer, cleanly, who owns each system, what data it uses, whether it is explainable, and where the human oversight sits.

How many boards actually use AI in their oversight? Recognition is rising faster than practice. PwC’s 2025 Annual Corporate Directors Survey found 35% of directors say their boards have incorporated AI into their oversight role, while Deloitte’s 2025 survey found 40% now think differently about board composition because of AI, but 66% still describe their boards as having limited to no AI knowledge. MSCI Institute found only about a quarter of large and mid-cap boards have even one director with AI expertise.

What are the biggest risks of using AI in the boardroom? Four stand out. Trusting a fluent but hallucinated output as fact; feeding confidential board information into an unapproved tool; creating a discoverable written record through speculative prompts; and letting AI-driven detail pull the board from oversight into management. Each is avoidable with an approved secure tool, a strict data line, a verification habit, and the discipline to use insight to ask better questions rather than to run the company.

Does a board need an AI expert director? It helps, but it is not a complete answer, and it can create false comfort. MSCI Institute found that even among boards with an AI expert, only a minority had used that expertise effectively. A single specialist does not discharge the whole board’s duty to understand AI risk well enough to oversee it. Building baseline competence across the board, so every director can ask the four oversight questions, usually matters more than one appointment.

What Indian regulations apply to a board overseeing AI? Several overlap. The Digital Personal Data Protection Act, 2023 governs personal data used in AI systems and carries penalties that reach the company. SEBI holds regulated entities responsible and liable for their AI outputs and investor-data integrity, with a board-approved governance framework among its expectations. The Companies Act, 2013 supplies the underlying duty of care and independent judgment, and any company offering AI-driven products or services in the European Union may also fall within the EU AI Act.

How do I write a good governance prompt for AI? Use four parts: role, context, task, and format. Give the model a role (“you are an experienced audit committee chair”), the real anonymised context and tension in the numbers, the specific task, and the shape you want the answer in, such as a ranked list of eight questions. Most weak outputs come from a one-line question with no context. A richer, governance-specific prompt produces a far sharper result, and using placeholders keeps confidential information out.

How long does it take a director to get competent with AI? About one board cycle of deliberate practice on real governance work. Set up the tool and your data and accountability lines in the first month, apply AI to your board-pack preparation and one pre-mortem in the second, and turn to overseeing the company’s AI in the third. Competence comes from reps on genuine work across a real quarter, not from a briefing. Persistence past the first attempt is what separates the directors who govern AI well from the ones who conclude it is overhyped.

References

Research and data

1. Governance of AI: A critical imperative for today’s boards, 2nd edition: Deloitte Global Boardroom Program, 2025 (survey of 700 board members and executives across 56 countries, January to February 2025)
2. 2025 Annual Corporate Directors Survey: PwC Governance Insights Center, 2025 (survey of US public company directors; 35% report AI incorporated into board oversight)
3. Enhancing AI governance on corporate boards: MSCI Institute (board AI-expertise analysis, data as of 30 June 2025)
4. Using AI in the Boardroom: New Opportunities and Challenges: Harvard Law School Forum on Corporate Governance, November 2025
5. 92% of Indian knowledge workers use AI in the workplace: 2024 Work Trend Index: Microsoft & LinkedIn, 2024 (India 92% vs global 75%)

Regulatory and legal

6. The Companies Act, 2013, Section 166 (Duties of directors): India Code, Government of India (Act No. 18 of 2013)
7. The Digital Personal Data Protection Act, 2023: Ministry of Electronics and Information Technology, Government of India
8. SEBI’s Proposed Amendments on Usage of AI Tools by Regulated Entities: Cyril Amarchand Mangaldas (analysis of SEBI proposal), December 2024

This article is for informational and educational purposes only and does not constitute professional, legal, financial, or governance advice. AI capabilities, board-oversight expectations, and the regulations governing AI are evolving quickly; verify the current position and consult a qualified professional before acting on any governance, disclosure, data-protection, or director-duty matter. Related reading: Duties and responsibilities of corporate boards (iPleaders).