Use of AI Technologies

Response

I am writing to request information under freedom of information legislation, regarding your organisation’s use of artificial intelligence (AI) technology/technologies. I would be grateful if you could provide the following detail: 

 

1) AI Systems in use

• A list of tools, platforms or systems currently deployed or being piloted/trialled.

• The purpose and function of each of the above.

• The departments or services where these are operational.

 

For the purposes of this request, Avon and Somerset Police has interpreted “AI technologies” as referring to standalone AI-enabled products and platforms, rather than every force system containing AI-assisted functionality.

 

Avon and Somerset Police have used the following tools relevant to this question since the date of your request:

 

– Microsoft Copilot Chat

– Söze (Soze)

– Pimloc SecureRedact (ceased being used 31 March 2026)

– NICE Investigate

 

A centrally recorded list of the departments or services using these products is not held. However, to assist, a brief explanation of the purpose and function of each product is provided below.

 

Microsoft Copilot Chat:

Copilot is a product used across a range of Microsoft products. Copilot Chat is available throughout Avon and Somerset Police and can be used in a number of different ways. This has been stated in an acceptable usage policy which is currently in draft state but is available for all users to view.

 

The licensed version of M365 Copilot is being trialled, however Avon and Somerset Police have not at this stage committed to any long term or permanent licenses.

 

Söze (Soze):

This is a product that has been used in proof-of-concept and proof-of-value contexts. It is an analysis platform capable of ingesting evidence and enhancing investigative capabilities through advanced pattern recognition, linking, search, et cetera, capabilities.

 

Pimloc SecureRedact (no longer in use):

This was a redaction solution which is billed by the vendor as being AI-powered, and having advanced capabilities for the search, review, and redaction of video media.

 

NICE Investigate:

NICE Investigate is the force’s Digital Evidence Management System. It is listed on the force intranet AI guidance page as an approved tool. The system provides a single view of digital evidence and includes functionality such as automated evidence collection, case-building and evidence management.

 

 

2) Procurement and development

• Details of any contracts, tenders or partnerships with external providers for AI solutions.

• Total expenditure on AI related technologies over the past three financial years, broken down by year.

 

The force does not record expenditure in a manner that identifies or categorises technologies as ‘AI-related’. Accordingly, no recorded breakdown of expenditure on AI-related technologies for the previous three financial years is held.

 

Copilot:

The Contract for Microsoft Copilot Chat is covered within a broader contract with Microsoft Enterprise Agreement. This includes other Microsoft packages and has a value of £9,436,156.92. This contract runs from 01/07/2025 through 30/06/2028. No specific contract cost for Copilot Chat is held.

 

Söze

The contract for Söze (Soze) Analytics Platform runs from 01/10/2024 through 30/09/2025 at a cost of £448,500.00.

 

Pimloc SecureRedact

No contract is recorded for Pimloc SecureRedact.

 

NICE Investigate

The force procured NICE Investigate under a Call-Off Contract with a total potential value of £2,696,487 over a 3+1 year term. This comprises:

• One-off implementation fee: £50,000

• Software-as-a-Service (SaaS): £1,640,000 over four years (including the optional extension year)

• Storage costs (estimated): £1,006,487 over four years

Annual charges are £410,000 per annum plus storage costs based on actual usage. Storage is charged quarterly in arrears at the following rates:

• Online GRS Storage: £35 per TB per month

• Offline GRS Storage: £5 per TB per month

As with the Microsoft Enterprise Agreement referred to above, this contract relates to the wider NICE Investigate platform and is not sufficiently granular to identify or attribute a specific contract value to any individual AI-enabled functionality within the system.

 

3) Governance and Ethical Oversight

• Copies of, or information relating to, any internal policies, frameworks or guidance documents relating to the use of AI.

• Any ethical review processes or risk assessments conducted prior to deployment.

• Details of any group responsible for the oversight of AI use within your organisation.

 

At the time of the request, Avon and Somerset Police held draft AI governance documentation but did not hold a finalised corporate AI policy. The following draft policies were held at the time of your request:

 

• Artificial Intelligence Policy

• Microsoft Copilot Chat Acceptable Use Policy

 

As these were still being formulated at the time of your request, I have not considered them to be internal policies within the scope of your request.

 

In relation to ethical review processes and risk assessments, Avon and Somerset Police undertakes governance and information assurance activity prior to the implementation of relevant technologies where appropriate. This may include completion of a Data Protection Impact Assessment (DPIA) and review through established information governance processes.

 

However, the Microsoft Copilot Chat Acceptable Use Policy has now been finalised. For transparency, given the delay in providing this response, I have provided a redacted copy of this policy outside of our obligations under the Act. Redactions have been applied to names, which would be exempt under Section 40(2) if provided under the Act.

 

The Artificial Intelligence Policy is still in draft format. It has therefore not been provided as it has not been considered within scope of your request, not being a finalised policy. We would however be able to consider a further request the draft policy as currently held. Our response would be subject to any applicable exemptions.

 

We also hold the following NPCC document available in the public domain via the link provided:

 

Responsible-AI-checklist-for-policing-2025.pdf

 

As this information is available in the public domain, it falls under the following exemption:

• Section 21 – information accessible via other means

 

Section 21 is an absolute and class based exemption and as such does not require a harm and public interest test.

 

This serves as a partial refusal notice under section 17(1) of the FOI Act.

 

 

4) Impact on Workforce

• Any assessments, reports or internal communications regarding the impact of AI on staffing levels, job roles or workforce planning (including recruitment, redundancy).

• Information on any roles that have been automated, restructured or made redundant due to AI implementation.

• Details of any training, redeployment or upskilling initiatives offered to staff in response to the adoption of AI.

•  Any consultations with trade unions or staff representatives regarding AI-related changes.

 

We have launched a pilot apprenticeship programme with a small cohort of officers and staff which is delivered by a third-party training provider. This course aims to upskill colleagues on the topic of AI generally.

In addition, Avon and Somerset Police has partnered with a third-party provider to establish an AI Transformation Academy for officers and staff. The academy offers a range of AI-focused development programmes aimed at improving AI literacy, leadership capability and the responsible adoption of AI within policing. Further information is provided in the attached leaflet.

Details of training relating to each tool/platform have been provided below. No further information is held in relation to this question. To confirm:

 

• No reports or assessments regarding the impact of AI on staffing levels have been produced

• No roles have been automated, restructured or made redundant due to AI implementation.

• No redeployment initiatives have been offered to staff in response to the adoption of AI, although training and upskilling opportunities have been made available through the AI Transformation Academy and system-specific training programmes.

• No consultations with trade unions or staff representatives have been recorded regarding AI-related changes

 

Details of training relating to Söze and Nice Investigate have been copied below, from the relevant pages on our intranet site available to all officers and staff. This includes details of upcoming training planned for September 2026.

 

Söze:

Training 

Training will be delivered in two stages to balance operational demand with the need to build capability:

• Stage one – delivered pre-summer demand, focussing on Sanctioning Officers and SPoCs, who will act as Söze Gatekeepers.

• Stage two – delivered post-summer demand, commencing from September 2026

 

 

NICE Investigate:

NICE Investigate training for core users

Around 3,500 officers and staff have been trained; the main training programme is complete. To request training contact Training (ICT).

Before you can start using NICE Investigate operationally you must complete two training elements. 

1. College Learn package  

The College Learn package must be completed at least 48 hours prior to your online training date. Failure to complete this training on time will result in you not being able to attend your scheduled session.

2. On-line training session 

• Your online training will be delivered via Teams and will be led by an ICT trainer with a maximum of 16 students.

• The training will be interactive and will require you to participate using the NICE Investigate software.

• The training duration will be approximately three hours including a break at the mid-point.

Training aims

The objectives of this training for attendees are to:

• Open and navigate the NICE Investigate system.

• Understand how Digital Evidence is uploaded to the NICE Investigate system.

• Search the NICE Investigate system to find evidence.

• Understand and use the functions available in the system.

 

While there are no specific training courses relating to Copilot and Pimloc Secure Redact, guidance and self learning for Copilot is available via our intranet pages and viva learn app. It is also possible that informal training has been provided within specific teams via team meetings. The force does not hold a centrally recorded repository of all informal team-based guidance or training relating to these systems. To determine whether such information is held would require enquiries across numerous teams and departments.

 

To the extent that recorded information may exist within individual teams, section 12(1) of the Freedom of Information Act applies, as the cost of determining, locating and retrieving such information would exceed the appropriate limit.

 

 

5) Performance and Evaluation

• Evaluations, audits or performance reviews of AI systems, as referenced in section 1.

• Evidence of how AI systems have affected service delivery, decision-making or operational efficiency.

Avon and Somerset hold no evaluations, audits or performance reviews of AI systems. The force also holds no recorded information assessing the impact of AI systems on service delivery, decision-making or operational efficiency.

 

 

6) Data protection and privacy

• Types of data used to train or operate AI systems, including whether this data is synthetic or not.

• Measures in place to ensure compliance with data protection legislation, including the DPA 2018 and UK GDPR.

• Procedures for handling bias, transparency and accountability in AI decision-making.

 

Avon and Somerset Police is not developing or training its own AI models and therefore holds no information regarding training datasets.

 

Compliance measures include completion of a DPIA and review through established information governance processes prior to deployment.

 

While some systems include AI-enabled or automated functionality to assist users, no AI system is currently used to make operational decisions without human involvement. As no AI systems are currently used for automated decision-making, the force holds no specific procedures relating to bias mitigation in automated decision-making.

 

Further information that may be of use is available on our website:

 

Our use of Artificial Intelligence | Avon and Somerset Police

 

In addition to the responses above, Avon and Somerset Police can neither confirm nor deny whether any additional information is held relating to the use of AI within Counter Terrorism, Serious and Organised Crime investigations, covert intelligence functions or specialist digital forensic capabilities supporting those functions.

 

The duty to confirm or deny is disapplied by virtue of the following exemptions:

 

Section 24(2) (national security)

Section 31(3) (law enforcement)

 

This response should not be taken as an indication of whether any additional information is held.

 

Confirming or denying whether AI is used by Avon and Somerset Police in counter‑terrorism, serious and organised crime, covert intelligence or specialist digital forensics would enable hostile actors to infer the presence, absence or maturity of particular policing capabilities. This type of capability insight is of practical value to adversaries who routinely adapt their behaviour to exploit perceived weaknesses, overwhelm less technologically equipped forces, or adjust methods to avoid detection. Even a single confirmation or denial could be combined with publicly available information from other forces to build a detailed national picture of capability. This mosaic effect represents a realistic and actionable risk, providing information that could materially undermine national security and the effectiveness of law‑enforcement operations.

 

Public Interest Test – Section 24(2) (National Security)

 

Factors in favour of confirming or denying 

There is a strong public interest in openness and transparency about the use of AI in policing. Confirming or denying the existence of information could contribute to informed public debate about how emerging technologies are deployed, the safeguards in place, and whether they align with legal and ethical standards. Disclosure can enhance accountability, promote trust, and reassure the public that AI use is subject to appropriate oversight.

 

Factors against confirming or denying 

There is an overriding public interest in protecting national security. Confirming or denying the existence of information about AI deployment in CT, SoC, covert intelligence, or linked specialist forensics would provide adversaries with useful insights into operational capabilities and priorities. Even a simple confirmation or denial could enable hostile actors to adjust methods, exploit potential gaps, or test thresholds. The resulting risk to national security would be real and more than trivial, and is not mitigated by partial or local confirmations, because the cumulative mosaic effect across multiple forces would still disclose sensitive capability at a national level.

 

Public Interest Test – Section 31(3) (Law Enforcement)

Factors in favour of confirming or denying 

There is a general public interest in understanding how AI is used to support policing functions, including the efficiency and effectiveness of core processes, and in ensuring that novel technologies are used lawfully and proportionately. Confirmation or denial could support external scrutiny and promote confidence in policing practice.

 

Factors against confirming or denying 

Confirming or denying the existence of information relating to AI use in CT, SoC, covert intelligence, or specialist digital forensics would be likely to prejudice the prevention and detection of crime, the apprehension or prosecution of offenders, and the operation of law enforcement tactics and methodologies. Knowledge of whether specific capabilities exist, or do not exist, enables offenders to adapt behaviour, exploit capability gaps, frustrate investigations, and increase the cost and complexity of counter‑measures. The public interest strongly favours maintaining an NCND to preserve the effectiveness of law enforcement and the safety of the public.

 

Balance Test

For the reasons set out above, Avon and Somerset Police will neither confirm nor deny whether it holds information indicating AI is used in CT, SoC, covert intelligence activity, or specialist digital forensics supporting those functions.

 

The public interest in maintaining the NCND position under sections 24(2) and 31(3) outweighs the public interest in confirmation or denial. This response should not be taken as an indication of whether any relevant information is held.

 

A separate response has already provided high‑level information about routine, non‑sensitive AI use. To go further would risk disclosing sensitive capability by inference.

 

 

If you are dissatisfied with this response, you have the right to request an Internal Review. Details of our Internal Review procedure can be found here:

 

Freedom of Information internal review procedure | Avon and Somerset Police