A Safer Future for Women and Girls: Investigation
Part 5
Part 5
Traditional investigative processes are being replaced by evidence-led models and advanced technology that deliver quicker, more reliable outcomes. Innovation is improving how cases are built and how victims are supported, with tools that accelerate analysis and strengthen decision-making.
Operation Soteria found several key high-level findings relating to the investigation of rape34:
There was a lack of sufficient specialist knowledge about sexual offending.
There was a disproportionate investigation effort being put into testing the credibility of the victim rather than a thorough investigation of the suspect’s offending behaviours.
There is a direct link between officer burnout, lack of learning and development and the confidence of officers in whether they are using the right investigative strategies.
Strategic analysis for recording rapes is fundamental to understanding the offending contexts in investigation strategies.
A new handheld device allows officers to better assess victims’ injuries on the frontline. Using a technique called cross-polarisation, the device improves colour contrast and enhances the visibility on injuries that may not be clearly visible to the naked eye. This is particularly useful for bruising which appears on darker skin tones during the earlier stages of injury. This gives stronger evidence which investigators can use to hold perpetrators accountable39.
A pilot in South London used this tool on 33 occasions, with charges brought against 45% of cases.
The Met has now expanded this trial across custody suits, sexual assault referral centres and forensic teams.
Söze is an artificial intelligence powered digital forensics platform trialled by Avon and Somerset Police which can enable investigators to access, analyse and tag extensive volumes of complex data from call records and financial transactions to map data.
In evaluation Söze reviewed evidence from 27 complex cases in around 30 hours - work estimated to take 81 years if reviewed manually37. The system has been instrumental in managing different crime types, including human trafficking, child exploitation and homicide.
This Police STAR project piloted the use of Natural Language Processing, content and sentiment analysis, stylometry and network analysis to infer authorship and connect anonymous online profiles40. This new approach to uncover the real-life identifies of offenders circumvented the usually time-consuming approaches to online identify verification and potential high operational security of offenders. This is particularly useful when working with anonymised profiles in a complex online environment that can evade traditional investigative approaches.
The project is being developed further by Hertfordshire Constabulary, with the inclusion of an ethical governance framework to provide transparency on the use of AI.
Stalkerware is software installed in a victim’s device without their knowledge or consent to secretly monitor their activity – from web searches, text messages, geolocation, voice calls and more. One example of defence against this intrusive software is SpyGuard, a free and an opensource, network -based scanner that safely detects stalkerware by analysing a device’s WiFi traffic for indicators of compromise without alerting perpetrators38.
Deployed by shelters and law enforcement, it enables victim centred triage and device hygiene advice in tech-facilitated abuse investigations.
Live Facial Recognition is a tool which uses live footage from a Facial Recognition (FR) camera to immediately scan a face, identify its features and compare the face with those on a watch list. If a match isn’t made, the image from the FR camera is immediately deleted.
An alternative technology used in investigations is Retrospective FR which is used after an incident as part of an investigation, the images come from CCTV, mobile phone footage, etc.
In 2023, the National Physical Laboratory independently evaluated the equitability of Live FR Systems and showed no statistically significant bias across demographic groups35. Since 2023 the Metropolitan Police and South Wales Police have increased the number of LFR deployments. The most recent year of deployments in the Met saw 962 arrests recorded, 25% of which were related to offences involving Violence Against Women and Girls36.
National Police Chiefs’ Council, Operation Soteria, n.d., https://www.npcc.police.uk/our-work/violence-against-women-and-girls/operation-soteria/.
National Physical Laboratory (for the Metropolitan Police Service & South Wales Police), Facial Recognition Technology in Law Enforcement: Equitability Study, 2023, https://science.police.uk/site/assets/files/3396/frt-equitability-study_mar2023.pdf.
Metropolitan Police Service, Live Facial Recognition Annual Report 2025, 2025, https://www.met.police.uk/SysSiteAssets/media/downloads/force-content/met/advice/
lfr/other-lfr-documents/live-facial-recognition-annual-report-2025.pdf.
Sky News, AI Tool That Can Do ‘81 Years of Detective Work in 30 Hours’ Trialled by Police, 2024, https://news.sky.com/story/ai-tool-that
-can-do-81-years-of-detective-work-in-30-hours-trialled-by-police-13220891.
GitHub (SpyGuard), SpyGuard, n.d., https://github.com/SpyGuard.
BBC News, Met Uses New Tech to Record Bruising on Dark Skin, 2025, https://www.bbc.co.uk/news/articles/c3vddwrl63qo.
National Police Chiefs’ Council / Science & Technology in Policing (Pulse Profile Project), Unmasking & Linking Software Engine – Complete Report, 2025, https://science.police.uk/site/assets/
files/3795/pulse_profile_unmasking_linking_software_engine_-_complete.pdf.
