Primary Market Bulletin 67

In October 2026, we launched a new interface for the NSM, our public archive for market disclosures.  

The enhanced service has been developed through extensive user research and testing. It aims to help investors, issuers and other market participants access the information they need more efficiently. This supports confidence in UK capital markets and reinforces our commitment to transparency and informed decision-making.

The new NSM functionality is described in our user guide.

The enhanced NSM is the first milestone in a wider programme to modernise the publication of markets regulatory information. Over time, we will bring together datasets to provide one authoritative place for users to find, access and reuse information. This free, trusted and accessible service will preserve records permanently for the public to use.  It will be called ‘FCA Markets Information Resource & Archive’ (or FCA MIRA for short). It will include:

• The modernised NSM. 
• A modernised Official List (a beta version will be made available as an initial step). 
• For each issuer, a profile page which will bring together reference data about the issuer and its securities with its NSM disclosures and short positions data.  
• The provision of an Application Programming Interface (API) to allow users to access and download the data programmatically in bulk.  

We will provide further detail about these changes in a future PMB. 

Please email [email protected] for opportunities to take part in testing future enhancements.

Background

We undertook a review of a small sample of sponsors to develop a better understanding of how they are using AI in connection with sponsor services.

In this article we share anonymised observations to give sponsors a sense of current and emerging market practice.

We recognise that AI is an area of rapid evolution and that firms’ approaches to using AI will differ depending on their business, systems, resources and risk appetite. 

While the sponsors we reviewed were at different stages in using AI, they all described it as supporting rather than replacing human judgement. They emphasised that responsibility for the sponsor’s work remained with the sponsor team and that AI-generated outputs were subject to human review. 

Our work

We selected a small number of sponsors that had recently completed a sponsor service. We requested information about whether and how AI had been used in the sponsor service (the transactions). 

We met with the sponsors to explore the information they had provided in more depth. Several sponsors said they had used AI in limited circumstances in relation to the transactions, but that their approach to the use of AI had developed since. So we were also able to gain an understanding of their current and emerging practice. Some sponsors demonstrated tools or described specific use cases to help us understand how they were using AI.

What we found

Rapid evolution of AI use

Several sponsors explained that using AI was encouraged within their firms. It was clear that the firms’ use of AI had developed since completing the transactions, emphasising the fast pace of development. In some cases, the use cases we were shown related to wider investment banking or corporate finance advisory work rather than sponsor services.

All sponsors had access to enterprise AI tools

The tools included general-purpose AI tools integrated on an enterprise-wide basis into existing office software, internally configured AI agents and a specialist AI tool designed for financial services. Some sponsors had also developed or configured tools for particular tasks. For example, locating regulatory material or carrying out quality-assurance checks. 

Differing levels of AI exploration and adoption for sponsor work

All the selected firms had access to firm-approved AI enterprise tools. However, the scope and maturity of use varied considerably. 

One sponsor continued to make very limited use of AI tools in sponsor work. The available AI tools were used for high level administrative support only, such as improving the quality of drafting. This sponsor emphasised the importance of sponsor analysis and human review.

Other sponsors in our sample were using AI more broadly. In addition to using AI for administrative and productivity support, it was used for: 

• Background research. 
• Summarising documents. 
• Generating first drafts of documents which staff then reviewed and refined. 
• Due diligence work to review documents and identify inconsistencies, gaps in reasoning or points that might warrant further consideration. 

Some sponsors were developing their use of AI to both identify precedents and use them as the basis for first drafts of announcements. One sponsor occasionally used AI to refine drafts of responses to FCA readers’ comments and to help formulate initial drafts of more standard responses.

At the more developed end of our sample, one sponsor had used a specialist financial-services AI platform to assist in several sponsor workstreams. For the transaction we reviewed, these workstreams were working capital, financial position and prospects procedures and aspects of prospectus and eligibility letter drafting.

We saw some use of AI agents by sponsors, for example, to assist with the presentation of work performed by the deal team. However, no sponsor was using AI for fully automated processes.

We also saw differences within firms, with some employees using AI more regularly than others. Several sponsors explained that adoption was developing iteratively as staff became more familiar with available tools and identified suitable use cases. None of the sponsors in our review mandated the use of AI. 

The importance of appropriate human oversight

A consistent theme was the importance of human involvement. AI was seen as a tool to support, rather than replace, professional judgement and technical expertise. 

All sponsors recognised the limitations of AI, and some sponsors noted the risk of hallucination. They also gave specific examples of limitations or errors in an AI-assisted output. These included the omission of a key comparator that the sponsor would have expected to have included in an analysis of a company against its peers.

All those sponsors who used AI for more than high level administrative support told us that AI-generated outputs were reviewed, validated or amended by staff before being relied upon. Responsibility for the final work product remained with employees.

Some sponsors told us there were certain activities during a sponsor service that were not generally considered appropriate for AI-assisted support. One sponsor said these activities included complex negotiations, board-level strategic advice and definitive ethical or legal accountability.

We were pleased to see sponsors exercising judgement about whether AI was suitable for a particular task. One sponsor said that when deciding whether to use AI it considered whether AI was likely to add value and the extent of human review required. 

In one example, a sponsor had used an AI transcription feature to help draft a file note of an important internal sponsor meeting. The file note was prepared by reference to the transcript and notes taken by individuals attending the meeting. The decision to use the transcription function took account of the breadth and complexity of the discussion and the importance of maintaining a comprehensive and accurate record of matters considered.

Risk management and governance

We saw a range of approaches for how sponsors’ systems and controls reflected the risks of AI adoption. Some sponsors had dedicated firm-wide AI policies. Others applied wider information security, technology or risk management frameworks to AI use. Recognising the pace of change, one firm had developed a set of overarching principles for responsible use of AI. In certain cases, policies had been recently revised or remained in draft as firms developed their governance arrangements.

Controls included restricting access to approved enterprise tools including blocking or limiting access to consumer AI applications and explicitly requiring human review of AI outputs. One firm’s policy expressly made individual staff responsible for both the information input to an AI tool and the output generated.

All sponsors emphasised the importance of client confidentiality. They all explained to us that the use of embedded AI tools meant that their data remained within their control and ownership. It was not used to train or enrich the proprietary model. 

All sponsors reflected that the adoption of AI tools had not created any risks requiring changes to existing sponsor-specific governance arrangements. Some sponsors explained that committees would consider and challenge papers and analysis presented on their merits, regardless of whether AI had been used to support research, analysis or drafting for example.

We noted that some sponsors were thinking about other risks posed by AI and how they might be managed. For instance, some, but not all, sponsors were considering whether AI-specific provisions should be included in their engagement terms.

One sponsor also identified the potential for risks arising from third-party advisers’ use of AI. These included an adviser failing to identify an error in AI-assisted work and confidential information being shared with an AI service provider that failed to keep it confidential. 

Overall, we were encouraged that firms were mindful of AI-related risks within their sponsor operations. They acknowledged that firm-wide policies, oversight and risk management appeared to be supporting responsible adoption of the technology, rather than sponsor-specific policies or procedures. As firms explore the use of AI within their sponsor businesses, we encourage them to continue considering the risks and ensure the appropriate guardrails are in place. Our AI in financial services: good practice and areas for improvement publication contains more information on governance, systems and controls considerations.   

Record keeping

Sponsors described how they had retained records which demonstrated and enabled verification of the basis for material judgements made when providing the sponsor service. Approaches to retaining prompts and outputs differed across sponsors and appeared appropriate in the circumstances. Overall, we were pleased that sponsors had considered and were able to explain on a case-by-case basis whether prompts and outputs constituted records. One sponsor explained that an AI generated document which was used to stimulate a discussion was not a record whereas the note of the discussion was. 

Conclusion

Our review found cautious and, in most cases, developing use of AI across sponsor firms. 

Sponsors told us they are using AI in a range of ways that improves the quality and efficiency of their work. We support sponsors in reducing unnecessary friction and cost while maintaining appropriate standards of diligence and investor protection. This has the potential for reducing the cost of sponsor services and making our listing regime more attractive to a wider range of companies. We encourage sponsors to continue to exercise this judgment and to continue to assess whether their governance, controls, oversight and record-keeping arrangements remain appropriate as AI’s capabilities continue to develop. We are happy to discuss innovative approaches with sponsors as they emerge. 

Performing the sponsor role involves suitably experienced professionals applying judgement to the information before them. AI may make that process more efficient and provide useful inputs, but it does not replace the need for the sponsor to reach its own conclusions after due and careful enquiry.

Critically, we were pleased that all sponsors considered how their use of AI fits with their responsibility to act with due care and skill. Sponsors consistently told us that AI may support sponsor work, but cannot replace a sponsor's professional judgement, independent challenge or accountability.

A new Update My Profile function is available in ESS. It enables users to request changes to their ESS profile, such as their name, job title, telephone number or email address, provided they remain with the same organisation. This removes the need to create a new registration when updating profile details and improves visibility of profile changes within ESS.

Using the Update My Profile function, users can submit a profile change request and provide a reason for the amendment. They may also choose whether approved changes should be applied to any relevant open cases. Users will receive email notifications when a request is submitted and again once the request has been approved or rejected.

Only one profile update request can be active at any one time.

Our ESS User Guide has been updated to reflect these functionalities and includes instructions on how to submit and track these requests.

Looking ahead

We will shortly introduce a new process for ESS users who move to a different organisation. This will make it simpler for existing ESS users to register at a new organisation, and will reduce administration and duplication within our systems. The process includes controls to ensure the prior organisation’s information will remain secure. 

We will provide further information ahead of this release.

Further information

See our updated ESS User Guide.

• Policy Statement PS25/9: New rules for the public offers and admissions to trading regime  
• PMB 63
• CP26/8: Quarterly consultation paper No. 51
• July 2026 Handbook Notice
• GC26/8: Primary Market Bulletin 67