Senior Company Secretaries and governance leaders came together to explore how directors are using AI in the boardroom, the opportunities it presents and the practical challenges of governing its adoption.
While organisations are progressing at different speeds, a consistent picture emerged: AI adoption is moving ahead of formal governance frameworks, and Company Secretaries are increasingly taking the lead in shaping how the technology is used responsibly and effectively.
Eight key themes emerged from the discussion:
The common thread: AI in the boardroom can no longer remain an entirely informal or experimental activity. The challenge is not simply how to introduce new technology, but how to ensure it strengthens the quality of information, independent challenge and board decision-making.
This paper explores the eight themes in greater depth and concludes with five questions for boards and governance teams to consider as they move towards more intentional AI adoption.
Imagine a director preparing for a board meeting. Faced with hundreds of pages of board papers, they turn to an AI tool and ask a simple question:
What is management not telling us?
The tool identifies an issue the director might otherwise have missed. At the next meeting, they ask a sharper question, challenge management's assumptions and potentially contribute to a better board decision.
Is that good governance? Or has the director potentially compromised the confidentiality of the board's information?
This tension framed the Global Governance Leaders Forum, which brought together senior Company Secretaries and governance leaders to explore the opportunities, risks and practical realities of AI in the boardroom.
The discussion built on insights from previous Global Directors Council roundtables, which highlighted three distinct conversations boards are having about AI:
While considerable attention has been given to the first two, the third remains relatively underdeveloped.
The Forum revealed that governance teams are increasingly finding themselves at the centre of this challenge, balancing the opportunities for innovation with the need to protect the confidentiality, independence and integrity of board processes.
The discussion explored how organisations are beginning to navigate these competing priorities in practice.
There is considerable variation in how organisations are approaching AI use by directors.
Some are already providing AI-enabled tools within secure board environments. Others are testing solutions internally, developing protocols or deliberately holding back until they have greater confidence in information security, legal privilege and data retention.
However, a consistent observation emerged: directors' enthusiasm for AI does not always translate into clarity about how they want to use it in the boardroom.
While directors increasingly recognise AI's potential, governance teams are often being asked to identify suitable applications, assess risks and establish the necessary safeguards.
This is creating a more proactive role for Company Secretaries. Rather than simply responding to requests for technology, they are helping determine where AI can genuinely improve board effectiveness.
Interestingly, the discussion challenged an assumption raised in previous director roundtables that adoption is often governed by the comfort level of the most cautious director.
Several governance leaders described a different reality, where the Company Secretariat, working alongside legal, risk, technology and information security functions, is increasingly driving decisions around AI use.
The takeaway: AI adoption in the boardroom is not simply a technology decision or a question of director appetite. It is increasingly a governance design question, requiring clear ownership, agreed protocols and ongoing oversight.
One of the clearest practical developments discussed was the growing interest in secure, organisation-controlled AI environments for directors.
Rather than allowing confidential board materials to be uploaded into publicly accessible AI tools, governance teams are exploring restricted environments where directors can interrogate approved information within established security boundaries.
These environments can provide access to board and committee papers, historical decisions and relevant organisational information, while maintaining greater control over what data is available and where it can travel.
However, placing AI inside a controlled environment does not automatically resolve every risk.
Participants highlighted several important questions:
The discussion also revealed that existing AI functionality within board portals does not always meet directors' expectations.
Basic summarisation may offer convenience, but it does not necessarily deliver the deeper insights directors are seeking.
This is prompting some governance teams to explore more tailored AI agents designed around their boards' specific needs.
The takeaway: The question is moving beyond whether directors should have access to AI. It is becoming how organisations can create an environment that is sufficiently secure, reliable and useful to support meaningful adoption.
Perhaps one of the most promising opportunities discussed was AI's ability to connect information across time.
Summarising a lengthy board pack is an obvious starting point. But several participants questioned whether summarisation alone delivers meaningful value, particularly when directors are still expected to read and critically assess the underlying material.
A more compelling application is the ability to interrogate years of board and committee papers.
For example, AI could help directors explore:
How has management's position on a particular risk evolved? What commitments were made several years ago? Which assumptions have changed? What was previously raised with the board but is absent from the current discussion?
This type of historical analysis has the potential to strengthen institutional memory, reveal inconsistencies and help directors ask more informed questions.
It could be particularly valuable for newly appointed directors, complex transformations, long-running strategic matters and recurring risk issues.
However, participants also recognised an important limitation. Historical board papers do not necessarily capture the full context of subsequent developments, informal discussions or decisions.
AI-generated findings therefore need to be tested against current information and validated through engagement with management.
The takeaway: The greater opportunity may not be helping directors read less, but helping them understand more. AI's value lies in connecting information, exposing gaps and strengthening challenge, rather than simply compressing content.
A recurring concern was the prospect of AI being used on both sides of the boardroom table.
Management may use AI to prepare or refine board papers. Directors may then use another AI tool to summarise those same papers and generate questions.
While both applications may improve efficiency, participants questioned what happens when human interpretation is progressively reduced.
If AI shapes the information presented and then influences how that information is interpreted, there is a risk that important nuance, independent thinking and constructive challenge are lost.
The concern is not necessarily that AI-generated information will be incorrect. It is that directors may increasingly rely on similar interpretations of the material, narrowing rather than broadening the diversity of perspectives around the table.
Participants reinforced that AI should support, not substitute for, directors' responsibilities to read, understand, question and exercise independent judgement.
An important distinction emerged between using AI to generate answers and using it to challenge assumptions.
A tool that simply recommends questions may encourage passive reliance. One that helps directors test alternative scenarios, interrogate assumptions or consider unfamiliar perspectives may contribute more meaningfully to board deliberations.
The takeaway: AI should strengthen independent judgement, not replace it. Its greatest contribution may be helping directors identify what they have not considered, rather than telling them what to think.
Few topics generated as much practical discussion as AI-assisted minute-taking, meeting transcription and recording.
There was broad recognition of the potential productivity benefits, particularly for governance teams managing substantial volumes of board and committee meetings.
However, approaches varied considerably.
Some participants described strict restrictions on recording or transcribing board meetings, citing concerns about confidentiality, legal privilege, discoverability and the creation of alternative records of board deliberations.
Others are already using AI to assist with minute preparation, drawing an important distinction between recording an entire meeting and processing notes independently prepared by a Company Secretary.
For example, some governance teams are using AI to transform their notes into draft minutes, improve structure and formatting, or incorporate relevant context from meeting papers.
In these cases, human review remains central to ensuring the final minutes accurately reflect the discussion, decisions and agreed actions.
A particularly important issue was the difference between deleting information and ensuring it is no longer retained.
Participants highlighted that deleting transcripts or prompts from a user-facing system does not necessarily mean the information has been permanently removed from underlying enterprise systems.
Retention policies, system logs, backups and organisational record-keeping requirements can all affect whether information remains accessible or discoverable.
This raised broader questions about whether governance teams have sufficient visibility into the technical infrastructure behind the tools they are using.
The discussion did not produce a universally accepted approach to AI-assisted minutes. Instead, it reinforced the importance of understanding precisely what information is captured, where it is stored, who can access it and what happens afterwards.
The takeaway: AI can offer meaningful efficiencies in minute preparation, but the distinction between drafting assistance and creating a separate record of board deliberations is critical. Any approach requires appropriate legal, technology and risk oversight, together with clear human accountability for the final record.
An encouraging theme was the extent to which governance teams are beginning to experiment with AI themselves.
Beyond board-level applications, participants described opportunities to improve the day-to-day effectiveness of Company Secretariat functions.
One practical example was using AI to review draft board papers before they reach directors.
By incorporating existing board paper guidelines, formatting requirements and expectations around content, AI agents can provide an initial assessment of whether a paper is sufficiently clear, complete and decision-focused.
This could include identifying missing information, inconsistencies, gaps in analysis or areas likely to attract challenge from directors.
Early experiences suggest these tools are far from perfect. Their outputs can lack sophistication, and the quality of results depends heavily on the information, context and instructions provided.
Nevertheless, they can serve as a useful additional layer of review, allowing governance teams to spend more time on substantive issues.
The discussion also explored the growing accessibility of AI agents.
Governance professionals without technical backgrounds are increasingly able to design tailored solutions using existing enterprise platforms.
These tools can support routine enquiries, information retrieval, document preparation and internal processes without necessarily requiring extensive software development.
This represents an important shift in the role of governance professionals.
The takeaway: Company Secretaries need not be passive recipients of AI technology. With appropriate support and safeguards, they can help design practical solutions that improve information quality, reduce administrative effort and strengthen the support provided to boards.
As AI becomes more important to organisational strategy and risk, questions around board composition and capability are becoming increasingly relevant.
Should AI expertise become a distinct requirement in the board skills matrix? Should boards actively recruit directors with specialist AI experience?
While the importance of technological literacy was widely recognised, participants were cautious about assuming that appointing a dedicated AI specialist would necessarily be the best solution.
One challenge is the pace of technological change.
Specialist knowledge can become outdated relatively quickly, particularly for non-executive directors who are no longer actively working in the field.
There was considerable interest in alternative approaches, including ongoing director education, access to independent specialist advisers and stronger technical expertise within executive management.
These approaches may allow boards to retain the broader commercial judgement and governance experience expected of directors while drawing on current specialist knowledge when necessary.
The conversation also extended beyond AI to other rapidly evolving areas, including cybersecurity and geopolitical risk.
The takeaway: The challenge may be less about securing a permanent AI expert at the board table and more about ensuring directors have access to the right expertise at the right time, while maintaining the capability to ask informed questions and exercise sound judgement.
Towards the end of the discussion, participants raised a more fundamental question.
If technology and business conditions are changing at an unprecedented pace, can boards continue to rely on largely static information prepared weeks before scheduled meetings?
The traditional cycle of preparing board papers, distributing a pack, holding a meeting and recording decisions has served boards for decades.
But AI creates opportunities to reconsider how information flows to directors and how boards monitor performance between meetings.
Could directors have access to more dynamic information rather than relying exclusively on fixed reporting periods?
Could AI help track commitments made to the board, identify emerging changes in risk or monitor progress against agreed priorities?
The discussion also extended to board effectiveness itself.
Annual board evaluations typically identify areas for improvement and generate agreed actions. Yet monitoring progress against those actions can remain episodic.
Participants explored the possibility of more continuous approaches, where boards and Chairs could assess progress against agreed objectives and behavioural improvements throughout the year.
These ideas remain largely exploratory and raise their own governance considerations, particularly around information quality, confidentiality and the appropriate boundaries of monitoring.
Nevertheless, they point towards an important longer-term opportunity.
The takeaway: AI may ultimately have a greater impact on how boards operate than on individual tasks within the existing board cycle. The opportunity is to move from periodic information consumption towards more continuous, informed and forward-looking governance.
The Forum highlighted a governance community that is neither universally resistant to AI nor rushing to adopt it without consideration.
Instead, organisations are progressing at different speeds, testing practical applications while working through legitimate concerns about confidentiality, legal privilege, information quality and director accountability.
There was no single agreed model for responsible boardroom AI adoption, nor a universally accepted position on areas such as transcription, retention or the level of AI access directors should have.
What emerged was a shared recognition that these questions can no longer remain entirely informal.
Five questions for boards and governance teams to consider:
1. Clarity of use: Do we understand how our directors are currently using AI, and have we agreed on appropriate boundaries?
2. Quality of information: Are we using AI simply to process information faster, or to improve the depth and quality of board challenge?
3. Accountability: Are directors and management maintaining sufficient independent judgement when AI influences how information is produced and interpreted?
4. Governance readiness: Do our policies, systems and retention practices reflect how AI is actually being used?
5. Future effectiveness: Could AI help us rethink how boards access information, monitor progress and make decisions, rather than simply automating existing processes?
Ultimately, the challenge is not choosing between innovation and good governance. It is ensuring that one strengthens the other.
For Company Secretaries and governance leaders, this represents an opportunity to move beyond managing the risks of AI adoption towards actively shaping how it can strengthen board effectiveness.
The measure of success will not simply be how much AI boards adopt, but whether it improves the quality of information, independent challenge and decision-making while preserving the judgement, independence and accountability at the heart of good governance.
Global Directors Council | Global Governance Leaders Forum
This paper captures themes and perspectives explored during a confidential peer discussion held under the Chatham House Rule. Insights are presented without attribution to individuals or organisations and are not intended to represent a consensus position or the policies and practices of any participating organisation.