A data subject access request that once ran to a single paragraph can now arrive as a ten-page letter, complete with detailed statutory citations, a prescribed search methodology and demands for “all metadata”, produced in minutes by a freely available AI tool.
New guidance from the Information Commissioner’s Office (ICO) on data subject access requests (DSARs) generated with AI, published on 11 November 2026, is its first dedicated guidance on the challenge for organisations in assessing and responding to the growing volume of such requests.
The headline message of the guidance was not unexpected: that a request is no less valid because it was drafted using AI. It also makes clear that length, poor drafting and legal citations do not, on their own, justify an extension of time for responding to the requester. Helpfully, however, the guidance confirms that controllers decide the scope of their own searches, need not engage with irrelevant legal references and may seek clarification where a request is genuinely unclear.
Key takeaways
- AI does not change the validity of a DSAR. The request must be assessed on its substance, not the method by which it was prepared.
- The controller determines the search. A requester’s list of systems, documents and metadata does not dictate the scope of a reasonable and proportionate search, but controllers must be able to explain their decisions.
- Clarification is not the same as complexity. A long or badly drafted request does not justify an extension — but where the controller genuinely needs further information to understand what is being sought, clarification may be the better route.
- Documentation is essential. Controllers should expect to be asked why they searched some systems and not others, why they extended the response deadline and what they did to verify the requester’s identity.
The guidance does not create new legal obligations. Rather, it explains how existing data protection principles and the amendments to the UK GDPR made by the Data (Use and Access) Act 2025 (DUAA) apply to AI-assisted requests. Two DUAA provisions are particularly relevant: (i) section 78 inserted Article 15(1A) into the UK GDPR, limiting the right of access to information that the controller can provide on the basis of a reasonable and proportionate search; and (ii) section 76 inserted a new Article 12A into the UK GDPR, which sets out the time limits for responding to data subject requests, including the ability to pause the response period where the controller reasonably requires further information to identify the information or processing activities to which a request relates. Section 76 came into force on 5 February 2026, while section 78 is treated as having had effect from 1 January 2024.
Finally, it is worth remembering that AI can help individuals exercise rights they might otherwise struggle to understand or pursue. As such, the guidance is not an invitation to treat AI-assisted requests with suspicion, but a reminder that controllers should apply the usual legal tests, regardless of how the request was prepared.
The controller — not the prompt — determines the search
AI-assisted requests often prescribe their own search methodology, including backups, named platforms, CCTV, call recordings and “all metadata”. The ICO confirms that requesters can identify where they believe their personal data are held, but it remains for the controller to determine the parameters of a reasonable and proportionate search.
There is, however, an important qualification: controllers should not simply ignore parts of an itemised request or make assumptions about what the requester wants. Rather, the ICO expects controllers to engage with requesters where appropriate and to document their search decisions, including the systems considered, the searches undertaken and the reasons for any exclusions. In practice, the distinction is between independently determining a reasonable search and disregarding inconvenient parts of the request. The former is the controller’s responsibility, whereas the latter can be more difficult to defend.
The guidance also draws a distinction between a request that is complex and one that is unclear. A request does not become complex simply because it is lengthy, repetitive or difficult to follow. But where the controller genuinely needs further information to identify what the requester is seeking, it may ask for clarification. This is particularly relevant where AI-assisted correspondence combines broad demands for personal data, metadata, particular search methodologies and material that may fall outside Article 15 of the UK GDPR.
Under Article 12A of the UK GDPR, the response period can be paused where the controller reasonably requires further information to identify the information or processing activities to which the request relates. The clock stops when clarification is requested and resumes when the necessary information is received. (The controller should only invoke this mechanism where further information is reasonably required to identify the information or processing activities sought, and should request clarification promptly.)
Where controllers are uncertain about what is being sought, they should consider whether clarification would be more appropriate than relying on the complexity extension. These are separate mechanisms, and the controller should be able to justify whichever approach it adopts.
The scope of the complexity extension
Under Article 12A of the UK GDPR, a controller may extend the response period by up to two further months where necessary, taking into account the complexity and number of the data subject’s requests. It must notify the requester within the initial one-month period and explain the reasons for the extension.
The ICO is clear that a request is not complex simply because it is long, convoluted, unclear or contains multiple legal references. Instead, complexity must arise from genuine difficulties in locating, reviewing or disclosing the information. Relevant factors may include the difficulty of extracting information, the need to consider exemptions and third-party rights, sensitive information, specialist advice or consultation with other organisations. The submission of multiple rights requests at the same time may also be relevant.
This is important because AI-assisted correspondence frequently combines access, erasure, objection and rectification requests with service complaints and other demands. The ICO accepts that multiple concurrent rights requests may justify an extension — although this does not mean that every request combining several rights automatically qualifies. For example, a service complaint, by itself, is not another data protection right and should not be treated as a complexity factor.
Metadata scope and limits
The ICO’s treatment of metadata is important, particularly for employment disputes. A generic request for “all my personal data, including metadata” does not necessarily require a search of every audit log or category of system-generated information. But a specific request, such as records showing who accessed the requester’s file, should be considered on its merits.
Metadata may constitute personal data where it relates to an identified or identifiable individual, whether directly or indirectly. Whether particular metadata falls within the scope of a SAR will depend on its content and circumstances. The ICO gives examples including access records, working patterns, call information and GPS tags in digital photographs. In employment cases, access logs, badge data and system activity records may be relevant to whistleblowing, discrimination and unfair dismissal claims, and employees and their advisers are increasingly identifying these categories expressly in their DSARs.
Accordingly, organisations should establish whether they can extract individual-level access and audit logs, how long retrieval takes and whether the records contain third-party information that will need to be protected. It is not enough to say that information is held in a technical log rather than in the substantive content of a document. Equally, not every system log will contain the requester’s personal data or fall within the scope of a reasonable and proportionate search. The assessment needs to be made by reference to the actual information and the circumstances of the request.
Security and fraud risks
While controllers are right to consider the broader challenges of responding to DSARs submitted using or generated by AI, they should also be alive to the associated security risks — namely, that AI can be used to generate inaccurate correspondence details, fabricated identity documents and convincing letters of authority. A realistic solicitor’s letterhead or letter of authority can now be produced with minimal effort, creating a risk both for organisations handling DSARs and for the individuals whose personal data may be wrongfully disclosed.
A response sent to an unauthorised recipient may constitute a personal data breach. Depending on the circumstances and the level of risk to individuals, this may trigger notification obligations under Articles 33 and 34 of the UK GDPR. It may also give rise to compensation claims under Article 82 of the UK GDPR and section 168 of the Data Protection Act 2018, where the relevant conditions are satisfied. For that reason, organisations should not rely exclusively on contact details or identity documents supplied with the request itself. Where a request involves a changed correspondence address, an unfamiliar representative or a questionable letter of authority, organisations should verify the request directly with the data subject, using contact details already held, before any personal data is disclosed.
Manifestly unfounded or excessive requests
The use of AI does not make a request manifestly unfounded or excessive, nor does the volume of requests alone. Nevertheless, the ICO recognises that AI may be used deliberately to generate disruptive requests and that the requester remains responsible for material generated on their behalf.
A clearly malicious, abusive or deliberately disruptive request may therefore support refusal, regardless of whether AI was involved in drafting it. This is relevant to organisations facing coordinated campaigns, including templated requests circulated on social media or in connection with employment and consumer disputes. Employers may also encounter repeated, substantially similar requests from a departing employee. In each case, any decision to rely on these provisions must be justified by reference to the individual circumstances of the request.
Moreover, the threshold remains high. Formal, assertive or aggressive language does not necessarily indicate malicious intent. Nor should a controller assume that a requester has an improper purpose simply because the request appears to have been prepared using AI. The ICO makes clear that controllers can explain where a request appears to misunderstand the law or seeks information beyond the scope of Article 15 of the UK GDPR, potentially because of the way it was generated. This is a useful practical point and reflects the fact that a controller need not spend time answering irrelevant legal citations or treating every demand in an AI-generated letter as a separate legal obligation.
Sectors and contexts most likely to be affected
Employment disputes
DSARs have long been used as a form of informal pre-action information gathering in employment matters. AI makes it easier to produce detailed requests identifying particular custodians, communications platforms, document categories and metadata. Requests may be timed to coincide with grievances, disciplinary proceedings, suspension, settlement negotiations or an anticipated tribunal claim.
The use of a DSAR in anticipation of litigation does not, of itself, undermine the requester’s right of access. Equally, the existence of an employment dispute does not turn a DSAR into a general disclosure exercise. Organisations need to distinguish between personal data that must be disclosed, information protected by legal professional privilege, third-party personal data and information potentially subject to other statutory exemptions. As an additional practical point, the DSAR should be handled under the organisation’s data protection framework but coordinated with the team managing the underlying dispute. Inconsistent positions taken by different teams about the same documents or events can create unnecessary difficulties.
Regulatory investigations and whistleblowing
Where an individual is a witness, complainant or subject in an internal or regulatory investigation, an AI-assisted DSAR may seek investigation materials, interview notes, internal correspondence and records of decision-making. The key distinction is between the individual’s personal data contained in investigation materials and the wider investigation documentation they may be seeking. Exemptions in Schedule 2 to the Data Protection Act 2018, including those relating to crime, regulatory functions and legal professional privilege, must be assessed against their specific statutory conditions, although they should not be treated as blanket grounds for withholding an entire investigation file.
High-volume consumer businesses
Retail, financial services, telecommunications and platform businesses are receiving more detailed and templated requests, and one operational challenge for these companies is to distinguish the access request from embedded complaints, other rights requests and demands for information outside Article 15 of the UK GDPR. A sensible triage process should identify the different elements of the correspondence at the outset and route them appropriately. It should not, however, become a reason to delay the DSAR while other issues are dealt with.
Next steps
The ICO’s guidance develops existing principles rather than departing from them, and confirms that AI changes how DSARs are prepared, not the legal framework that governs them. The important questions remain the same: what personal data are being sought, what is a reasonable and proportionate search, and can the controller explain the decisions it has taken? Organisations facing increasingly detailed and contentious DSARs should focus on the following:
- Review extension and clarification practice: (i) audit recent AI-assisted DSARs; (ii) establish whether extensions were justified by genuine complexity or multiple rights requests, and whether clarification would have been more appropriate; and (iii) stop any automatic extension practice based simply on the length or drafting of a request.
- Check identity and authority procedures: (i) review how the organisation deals with changed correspondence addresses, third-party representatives and letters of authority; and (ii) ensure that there is a practical means of contacting the data subject directly where authenticity is in doubt.
- Document search decisions: (i) introduce a short scoping record for non-routine DSARs, identifying the systems considered, searches undertaken, information excluded and reasons for those decisions; and (ii) record any engagement with the requester.
- Understand metadata capabilities: establish with IT what access logs, audit records and other metadata can be retrieved for an individual, how long retrieval takes and how third-party information would be protected.
- Involve the legal team where there is a dispute: (i) DSARs from current or former employees, complainants and counterparties in litigation or regulatory investigations should be flagged promptly; and (ii) the data protection team should coordinate with those managing the underlying matter, without confusing the DSAR process with wider disclosure obligations.
Subscribe to Ropes & Gray Viewpoints by topic here.
Authors
Stay Up To Date with Ropes & Gray
Ropes & Gray attorneys provide timely analysis on legal developments, court decisions and changes in legislation and regulations.
Stay in the loop with all things Ropes & Gray, and find out more about our people, culture, initiatives and everything that’s happening.
We regularly notify our clients and contacts of significant legal developments, news, webinars and teleconferences that affect their industries.
