AI Responsibility Statement
1. What AI does in Cindr
Cindr uses AI language models to help fire risk assessors turn rough site notes into structured report prose. The assessor types brief observations into a section's notes field. Cindr sends those notes to an AI model, which returns draft text written in plain, professional English consistent with PAS 79-1 and BS 9792:2025 reporting style.
AI is used for this text generation step, and to suggest remedial actions from the assessor’s notes. A suggested action only reaches the report if the assessor accepts it. AI is not involved in scoring risk, in deciding which findings are significant, or in any other part of the assessment process.
2. What AI does not do
Cindr's AI does none of the following.
- Conduct or supervise fire risk assessments
- Determine whether a finding is significant, low, medium, or high risk
- Decide remedial actions or timescales. It can suggest actions from the assessor’s notes, and the assessor accepts, edits or rejects each one before it enters the report
- Verify the accuracy of information provided by the assessor
- Replace the knowledge, judgement, or on-site expertise of a qualified fire risk assessor
AI-generated text is a drafting aid. It reflects only what the assessor has chosen to input. It cannot observe the premises, identify hazards, or apply professional discretion.
3. Assessor responsibility
All report content produced using Cindr remains the professional responsibility of the named assessor. Before any report is issued to a client or responsible person, the assessor must complete all of the following.
- Read and review all AI-generated sections in full
- Correct any inaccuracies, omissions, or language that does not reflect the actual site conditions
- Ensure the completed report meets the requirements of PAS 79-1 or BS 9792:2025 as applicable, and any relevant accreditation body (such as BAFE SP205)
- Apply their professional judgement to all findings, risk ratings, and recommended actions
- Accept, edit or reject each suggested remedial action
Cindr accepts no responsibility for the content of reports issued to clients. The assessor's professional judgement is final.
4. Known limitations of AI-generated text
AI language models can produce text that sounds plausible but is factually incorrect. In a fire safety context this is a serious risk. Common failure modes are listed below.
- Misinterpreting abbreviated or ambiguous notes
- Generating generic language that does not accurately describe the specific premises
- Producing text that omits important caveats present in the assessor's notes
- Using terminology that does not align with current PAS 79-1, BS 9792:2025 or BS 9999 language
For this reason, every AI-generated output in Cindr is presented as editable draft text, not as a finished finding. Assessors should treat it as a first draft and edit it to reflect the actual situation on site.
5. Data sent to AI services
When an assessor generates a report section, the text entered in the site notes field is sent to a third-party AI provider to generate the draft prose. We do not send personally identifiable information about building occupants or responsible persons to AI services. Assessors should not include names, contact details, or other personal data in the notes field.
Details of the specific AI provider used and how data is handled are available on request at info@cindrapp.com.
6. Feedback and quality
We actively monitor the quality of AI-generated outputs. If you encounter text that is inaccurate, misleading, or otherwise unsuitable for inclusion in a professional fire risk assessment report, please let us know at info@cindrapp.com. Your feedback directly informs how we improve the system.
7. Changes to this statement
As AI capabilities and our use of them evolve, we will update this statement. Material changes will be communicated to account holders by email.