Can AI Help Solve the Disclosure Crisis?
The UK’s disclosure crisis has been a headline story for more than two years, piling pressure on police and prosecutors. The impact on justice is stark: failures in disclosure can enable the guilty to walk free, whilst also risking the wrongful conviction of the innocent.
Police and prosecutors have a legal duty to disclose to the defense material or evidence unearthed that they will not be using and which may establish the innocence of the accused. But according to new research by The Times, an average of 120 criminal cases collapse every month because of failures to disclose evidence.
There are many causes of the disclosure crisis, but most commentators agree that the deluge of digital data – combined with ever-more limited resources to process it – is chief amongst these.
When I became a police officer thirty years ago, the paper evidence collected in a typical case would have involved everything from witness statements to the suspect’s fingerprints and the handwritten notes officers had taken in the course of the investigation. Police still need to collect all this information– but today there is also a digital evidence trail that needs to be accessed then downloaded, decoded and investigated.
A single average-sized smartphone contains 50,000 pages of data, so it is easy to see how the same kinds of cases I attended to in the 1990s could now generate hundreds of thousands of pages of evidence – not to mention the data generated by CCTV and ANPR systems, satellite navigation systems, fitness watches and the myriad other devices that have become ubiquitous in modern life. And as more and more mobile phones and apps become encrypted, the data is becoming increasingly complex to decode and process.
Combine the proliferation of digital evidence with the constrained resources faced by police and prosecutors, and you have a recipe for many miscarriages of justice. The data generated during investigations will only increase in volume and complexity as phones grow in memory and new types of devices and technology grow in popularity.
Cellebrite built its reputation by working with police forces first to lawfully access the devices of suspects, then to extract and decode the data. Cellebrite Pathfinder delivers the next stage of this process and breaks new ground in the use of Artificial Intelligence and machine learning in police investigations.
Through this tool, which automates the review of digital evidence including video footage, police forces in the UK and around the world are being enabled to crunch through enormous volumes of evidence faster than they have ever been able to before, with minimal resources. And then to present the data in the most meaningful format for the CPS, judge and jury.
The impact this tool can have upon the most complex and significant investigations is immense. The North West Counter Terrorism Unit, for example, has been using Cellebrite Pathfinder as it investigates the Manchester Arena terror attack.
By drastically speeding up the analysis of digital evidence, the tool is already credited with putting behind bars criminals who may otherwise still be walking the streets, including this car theft gang in Gloucester.
At the time, the Detective Sergeant in charge, Harry Limer, explained the difference our tool made to his investigation: ‘…to do the procedures manually, it would have taken months…now instead of police officers working long hours for months, while having to increase our budgets for overtime, we were able to leave the software running overnight to get results the following morning.’
Fundamentally, the tool surfaces new leads and reduces the time to evidence, enabling police forces to solve more crimes, faster – while drastically simplifying disclosure in the face of reams of digital evidence. It:
- Eliminates the time-intensive manual review of digital data often running into hundreds of thousands of pages in a single investigation;
- Enriches investigations by merging different data sources from the multiple aliases a suspect might have into a single profile – and identifying who they have been talking to (and when) or meeting with (and where);
- Allows police to find, in a few clicks of their mouse, images (including from video footage such as CCTV) relating to 15 media categories such as drugs, money and weapons – but they can also train the machine to recognise and find new categories without having to wade through the thousands of images (and thousands of hours of CCTV footage) that would typically exist in a case;
- Can work in any language, eliminating the need for translators and accelerating timelines in multi-language cases.
There is no quick-fix solution to the disclosure crisis, but just as technology helped to create the problem, it also creates an answer.