We are currently experiencing network problems with the desktop version of Police Oracle. We hope to have these resolved as soon as possible.

Automation and AI within CSE

Following a successful pilot last year, GMP are planning on automating processes for identifying indecent images from January.

From the new year, GMP are set to automate the initial processing for indecent images. 

It follows a successful pilot driven by the Transforming Forensics programme within the FCN. 

The initiative centres around technology developed by Canadian company Magnet Forensics which enables the creation of workflows to automate initial processing of exhibits. 

The tech is capable of doing the initial imaging and sorting data into types, as well as running datasets past key words and hash sets. 

The system can run 24/7 and GMP currently have six workstations that they will kickstart from January. The pilot initially focussed on low risk, no contact cases. 

Initial results based on a small sample size have indicated cases were processed 55% faster and an average of 9.5 hours were saved per case. It’s also reduced the human exposure to indecent images. 

GMP digital forensic quality and development co-ordinator Chris Whiteley told Police Oracle: “[Everyone’s] seen recently the statistic of the 25,000 devices backlog. GMP are in the same position, we're talking up to 12 months before we can even get around to some of these jobs- which is scary really. 

“There could be jobs in there that really do need to be pushed up that risk list and prioritised and you just don’t know about it. 

“Eventually, when that job comes to the point where you can get the exhibits called in and do the imaging - it might sit for another five to six weeks before we can get an investigator to pick it up and then start doing all that processing work. 

“Some of these jobs could be quite quick and easy to get rid of. Some of it could involve the police officers more in terms of reviewing, and some of them actually do need to be pushed up the queue and dealt with asap.” 

Putting exhibits through automation processing possibly means a quicker identification of risk. Mr Whiteley gives the example of a one image upload. 

“It could be a kind of a character - a cartoon or something or it could be a child suspect who thinks it's funny. Is it right for them to be sat in that queue for a year? We could get that job through and find actually out that if that’s all it is then have a chat to a police officer and job done. 

“It could turn out we've got thousands of indecent images downloaded, in which case we'd need to progress that a bit quicker. It's more about taking in jobs and getting them done faster. Even if they are sitting waiting to be investigated, we can still prioritise them and know where the risk is.” 

A possible future avenue for this is the additional use of AI. It’s something GMP have piloted previously but are looking to explore further in the future. 

AI could identify percentage likelihoods of an image containing certain features- skin tones for example. It would help for images that aren’t already in CADE or ones which have been adapted so that it’s not recognised by the database but still depicts the same thing. 

Police Oracle asked whether there was a risk involved with this around relying on tech and therefore possibly missing things. 

Mr Whiteley explained: “Mobile phones they’re the biggest challenge in this area. Apps are changing all the time. One minute, you can get into this flow, and the next minute they move it and put it somewhere else.

“There's always always a chance that there could be something missing. But I think the way we've got to look at these jobs is that it's still going to go to an investigator, they're still going to do their job. Just because we’ve done some pre-processing doesn't necessarily mean we can't do anymore. 

“[The investigators are in control of their investigations] but if we can reduce the work in this way, it frees up officers for the more complex work.” 

While the bulk of the backlog remains in child exploitation and it’s here taht GMP will focus initially, Mr Whiteley said the tech could be expanded for use elsewhere in the future - pulling word docs and Excel spreadsheets in a fraud case for example. 

An on-going challenge is getting the data to the relevant officers. Currently the force are using discs and USBs - ideally Mr Whiteley would like a secure online review platform. 

Earlier in the year, John Beckwith, TF’s DF Capability Lead, said of the project: “We’re extremely proud of what we’ve delivered with forces on CSE Automate. Early results around the potential impact of automation on case processing times and staff wellbeing are exceptional and hold huge promise.

“Together, we’ve created something which has the potential to be a genuine game changer for policing and, most importantly, to the victims of CSEA and other crimes”

Leave a Comment
In Other News
PSNI steps up targeted searches to crack down on CSE
New NCA technology helps jail prolific child rapist
Cheshire’s IT team: “There isn't any more we need to buy”
Digital forensics inspection 'paints a sorry picture' for policing
Project Odyssey: a breakthrough in sexual abuse evidence gathering
'Industrial scale' CSA work being mopped up by non-specialists
More News