A Safer Future for Women and Girls: Forensics

Forensics is now about speed, precision, and protection. From early digital tools to cutting-edge AI that can analyse injuries and detect abusive language, innovation is transforming how VAWG cases are investigated. Advances in mobile and real-time forensic technology also now enable faster, more accurate evidence collection. Policing have also been grappling with the challenge of explosion in digital material – over 90% of recorded crime now has a digital element.

2023

Deepfake Detection

National

Deepfakes are a form of synthetic media generated or manipulated by AI that can focus on manipulating visual or auditory information to create convincing fake sexual content. A recent survey into public attitudes on deepfakes revealed that 92% of the 1,700 participants aged 16 and over are concerned about sexual deepfakes. Worryingly, the survey also revealed that 25% of people either agreed or feel neutral about the viewing, creating or selling an intimate non-consensual deepfake²⁶.

Humans on average, can only detect 63.3% of deepfakes accurately. Policing, the Home Office and other government bodies are collectively working at pace to effectively detect deepfakes using artificial intelligence tools with super pattern-spotting functionality. The 2025 and 2026 Deepfake Detection Challenges^27,28, spearheaded by the Home Office, have brought together industry and academia to produce solutions to detecting deepfakes.

2023

Natural Language Processing Threat Detection Platform

National

The Forensic Capability Network and University of Warwick collaborated on a Natural Language Based platform that significantly enhances the effectiveness of handling digital text evidence to detect sexist threats and abuse online²⁹. The platform enabled rapid categorisation of messages and delivered contextual summaries and persona profiling based on conversational data. 

In one test, the model identified abusive language around 21 times faster than an average human investigator³⁰.

2023

DigiVans

Bedfordshire

The DigiVan is a fully mobile, unmarked digital forensic unit which provides officers and forensic staff with faster and easier access to examine digital devices at any location and enables them to be returned immediately after they have been processed.

The team acquires as much evidence as possible at the scene and hands back devices to the victims, offenders, or witnesses at the scene, wherever possible. This approach ensures that in many more cases we can return devices to people within 2 hours. In just one year, Bedfordshire’s Digital Forensics Unit (DFU) backlog for computers and phones dropped from 471 to 38 days, and there has been an 80% reduction in turnaround times for mobiles, as well as a 90% reduction in backlogs of devices to be examined.

2024

AI Tool for forensic investigations of traumatic brain injuries

Oxford

There is currently no standardised and quantified approach to determining whether physical impact can cause a reported traumatic brain injury (TBI). A study from Oxford University sought to use an AI framework based on forensic data from the National Crime Agency’s National Injury Database and 53 anonymised police reports from Thames Valley Police to predict TBI scenarios based on documented assault scenarios. The biomechanical predictions of strain and stress distributions in an assault and specific assault metadata are interpreted by a machine learning model to predict brain injury outcomes.

The model achieved the following³¹:

94% accuracy for skull fractures

79% accuracy for loss of consciousness

79% accuracy for intracranial haemorrhage (bleeding within the skull)

2025

Forensic Marking for Domestic Abuse Perpetrators

College of Policing

Forensic Marking uses a unique, chemically marked grease or water-based solution which is typically used to deter property theft. The College of Policing extended the use of forensic marking into a domestic abuse context to improve the detection of perpetrators interacting with their victims and the deterrence of perpetrators from approaching victims. The evaluation of this new technique showed promising results; there was a consistent increase in feelings of safety and freedom for victims and forensic marking was associated with a 22% reduction in repeat incidents over a 6 month follow up period³².

2025

Y- STR DNA Profiling 

National

To tackle sexual assault, the UK Government has committed £7 million to the development of a Y-STR profile database³³. Y-STR DNA strands are specifically from the Y-chromosome, present only in those born genetically male. A complete database will enable policing to match male-specific DNA with those from crime scene samples and potentially speed up investigations and open new lines of enquiry where the offender would be otherwise unidentified.