AI Generated Complaints

AI generated complaints are becoming increasingly common across the public sector. This resource offers practical guidance to help general practices respond appropriately, proportionately and confidently.

Date published:
Date last updated:
Next review date:

AI generated complaints are becoming increasingly common across the public sector. Alongside numerous other Local Medical Committees, Surrey and Sussex LMCs has observed an increase in complaints received by practices that appear to have been written with the assistance of, or in some cases entirely by, generative artificial intelligence (AI).

This trend has also been identified across the public sector, with bodies such as schools, local government, NHS Trusts, and even the financial ombudsman describing an increasing number of complaints which apparently include AI assisted text. This is thought to be contributing to an escalating number of received complaints. Interestingly, the GMC also noted a 25% increase in complaints about doctors in 2025 compared with 2024, although they did not comment in terms of any AI generated content.

The use of AI does not invalidate a complaint and should not alter how practices approach complaints management. However, AI generated correspondence can create practical challenges in terms of workload, clarity and the production of proportionate responses.

Why can AI generated complaints be challenging?

Practices report several recurring issues when reviewing complaints that appear to have been generated or expanded using AI.

  • Longer and More Complex Correspondence
    AI generated complaints are often significantly longer than traditional complaints. They may contain repetition, multiple overlapping concerns and extensive background information, making them more time consuming to review and understand.
  • Longer and More Complex Correspondence
    AI generated complaints are often significantly longer than traditional complaints. They may contain repetition, multiple overlapping concerns and extensive background information, making them more time consuming to review and understand.
  • May use Formal or Adversarial Language
    AI generated text can sometimes appear excessively formal and antagonistic, including disproportionate claims of harm, threats towards staff involved, or reference to legal penalties which again may not be relevant.
  • Risk of Misrepresenting the Patient’s Views
    Patients may place considerable trust in AI generated content and may not always review it carefully before sending it. As a result, the final correspondence may not accurately reflect what the patient intended to say or the outcome they are seeking.

Although receiving apparently AI generated complaints may cause concern, it is important to treat such complaints in the same way as others, particularly in terms of initial acknowledgement. There is no requirement for patients to make an informal or verbal complaint before contacting the practice although a practice can encourage patients to do so if it would assist in resolving a complaint.

It is also important not to overtly criticise the use of AI, as it can be a valuable aid, particularly for patients who may struggle to articulate what they want to say clearly or have the confidence to express their position in writing. Consider the context of the complaint in terms of patients or relatives who may perceive the NHS is simply not functioning efficiently and delivering the care they expect. That does not mean your practice has to do the impossible or be a substitute for other responsible organisations, but acknowledging this perception and the limits of your practice in terms of delivery of care, without criticising other organisations, is reasonable.

What if the complaint is difficult to understand?

Unfortunately, there may be no way round initially reading the complaint in full, which may represent a significant workload for practices. In some cases, the complaint may be so broad that it is difficult to determine precisely which issues the patient wishes to have investigated. Where this is the case, it may be appropriate to ask the patient to clarify their concerns.

The LMC does not recommend referring to any actual or suspected use of AI, or making negative comments about AI-generated submissions. Instead, practices may wish to invite the patient to clarify the matters they would like investigated, explaining that this may assist in providing a timely and meaningful response to their complaint.

Patients can be asked to describe their personal experience, in their own words, as this is the most valuable way of managing any complaint, including:

  • What happened.
  • When and where did it happen.
  • Who was involved, including both staff and other patient family members.
  • How did it affect you or your family.
  • What outcome do you personally want to see.

This approach may create a more easily understood complaint; later in this letter is a sample advice sheet that may be included in a ‘complaints pack’.

When complaints are shared with other organisations

Some patients may copy their complaint to others, including NHS England, the GMC or their local MP. This may be the case whether or not AI is used to generate a complaint. This is not recommended as it may create confusion in terms of responsibility for managing a complaint, although organisations such as NHS England and the GMC should advise patients that they will normally only become involved if a complaint cannot be satisfactorily resolved at a practice level.

It is important not to overtly criticise the use of AI, as it can be a valuable aid, particularly for patients who may struggle to articulate what they want to say clearly or have the confidence to express their position in writing. It is also important to consider the context of the complaint in terms of patients or relatives who may perceive the NHS is simply not functioning efficiently and delivering the care they expect. That does not mean your practice has to do the impossible or be a substitute for other responsible organisations, but acknowledging this perception and the limits of your practice in terms of delivery of care, without criticising other organisations, is reasonable. Practices should also not input a patient’s personal identifiable details into any generative AI system unless their privacy can be assured; the LMC has included the same advice in the accompanying template letter for patients.

What about using AI within the Practice

The LMC also advises caution regarding a practice’s own use of AI when reviewing complaints or drafting responses.

AI systems frequently adopt the tone and structure of the material they are analysing. Where the original complaint is lengthy, repetitive or adversarial, the resulting draft response may reflect those characteristics.

If practices choose to use AI in patient communications, they should ensure that:

  • All letters are carefully reviewed, particularly in terms of factual information received from the patient, or in reference to facts the practice relies on. This takes experience and a familiarity with the patient’s original complaint.
  • Keep the letter focussed on the matter(s) in hand, raised within the patient’s complaint. Remove any irrelevant or repetitive information or phrasing.
  • Ensure the letter reflects what the practice wishes to say in an empathetic and reassuring style.
  • Read and review the final draft version.

FAQs

Yes. The important issue is the substance of the complaint rather than the method used to draft it. Practices should avoid criticising patients or making assumptions about why AI may have been used.

Not necessarily. The complaint should be managed through the practice’s normal complaints process. The focus should remain on identifying the underlying concerns, seeking clarification where necessary and providing a proportionate response.

Not necessarily. Practices may find it more helpful to group related issues into themes, provided the substantive concerns are addressed clearly and no distinct issue is overlooked.

This will not usually be necessary. If a complaint is unclear, it is generally more helpful to ask the patient to clarify their concerns or the outcomes they are seeking.

Practices may summarise the key issues or themes they have identified. If however, the complaint remains unclear, then you could invite the patient to describe their personal experience, in their own words, for example:

  • What happened.
  • When and where did it happen.
  • Who was involved, including both staff and other patient family members.
  • How did it affect you or your family.
  • What outcome do you personally want to see.

Practices should check whether references are accurate, current and relevant. The presence of technical, legal or regulatory language should not be treated as evidence that the cited material applies to the circumstances being complained about.

Potentially, but only in accordance with the practice’s information governance requirements. Patient identifiable or confidential information should not be entered into AI systems that have not been appropriately approved for that purpose.

No. Patients may complain verbally or in writing and are not required to attempt informal resolution before submitting a complaint.

No. The use of AI is not, in itself, grounds for removal. Any decision relating to behavioural concerns or relationship breakdown should be based on evidence and considered separately from the complaint itself.

Practices may wish to consider whether any new information has been provided, whether the local complaints process has concluded and whether a further substantive response is required. Clear communication and reasonable boundaries may help avoid unnecessary duplication.