Serious Fraud Office uses artificial intelligence to crack real crimes
Chief technology officer Ben Denison discusses how the organisation is using technology to get on top of increasingly vast and complex cases of bribery, fraud, and corruption
The Serious Fraud Office is a specialist prosecuting authority tackling the top level of serious or complex fraud, bribery, and corruption across England, Wales and Northern Ireland.
It both investigates and prosecutes its cases, which is unique but necessary because the cases are complicated, and lawyers and investigators need to work together from the outset of them.
Supporting both lawyers and investigators is an IT infrastructure that is managed by chief technology officer (CTO) Ben Denison, who also looks after the technology that supports the SFO’s operational activities.
“That includes digital evidence, so in our cases when we’re investigating someone we’ll see information from phones, tablets, laptops, and email, and we have a digital forensics team that processes that material and extracts relevant data from it, and it can then be ingested into another system which our case teams use to review it from,” Denison tells PublicTechnology.
In essence, the SFO ensures that all of the data – whether it’s digital or a hard copy – is put into one place, so that it can be reviewed together. But the scale of this data is growing by the day, making it a challenge for the organisation.
“Our cases are getting much bigger; up until last year the biggest case we’ve worked on was the Rolls-Royce case which had 31 million documents, and now we have another case that is 50 million documents, and we’re getting to the point where you have to use technology to review this material, otherwise it would take forever to go through each document,” Denison says.
So, the organisation is looking at using technology to reduce the amount of materials needed to get the most relevant parts of the data. An example of this in action is the organisation’s use of artificial intelligence to trawl through documents to check which materials are covered by legal professional privilege – as these cannot be accessed by the SFO, or used for any investigation.
Number of documents process by the SFO’s digital forensics team each month as of the start of 2018
Amount of times faster than a human lawyer the SFO’s AI system can process documents
Settlement paid by Rolls-Royce in 2017 bribery case
Number of documents processed by the SFO during the Rolls-Royce case – which, at the time, was its most complex ever
Previously, independent barristers were tasked with reviewing this material.
In the past, they searched for relevant words within the materials, but this was returning a large number of positives – and many of the materials were not actually privileged.
“We’re now able to use AI technology to reduce the number of privileged documents by a significant amount, meaning we’re able to spend a lot less time having to review that material because we were able to weed out what definitely wasn’t covered by privilege at a much earlier stage,” Denison says.
OpenText’s Axcelerate is the AI-powered document review system that the SFO used, initially as a pilot for the Rolls-Royce bribery case, which was settled for £671m earlier this year.
The system has enabled the organisation to save about 80% of costs and time in that part of the Rolls-Royce investigation, Denison says, with the system able to process more than half a million documents a day at speeds 2,000 times faster than a human lawyer.
Now, the SFO is looking to roll that technology out for all of its cases, and combine it with other analytical tools to identify key information about people, recognise patterns, group information by subject, organise timelines and remove duplicates. This will all reduce the time it takes for the SFO to get to trial.
Tech influence in serious fraud
Fraud in general is being impacted by the rise of easy-to-use software and techniques to con people to give up details or transfer money through methods such as phishing emails. These methods are getting easier for criminals to use and harder to thwart. However, as the sophistication of the technology used in this kind of fraud is on the rise, it’s not necessarily the case for the types of serious cases that the SFO deals with.
“We’re dealing with top-tier cases, and we only take on a small number of cases every year and those tend to involve similar sorts of crime – that’s not to say technology isn’t used as a part of them, but it’s not used in the same way as in those low-level high-volume fraud cases,” Denison says.
There are many tools available on the market to analyse large volumes of visual material, but again this does not apply to the types of cases the SFO deals with.
“We’re not using CCTV evidence or dealing with large volumes of visual material, although we do have it as part of our evidential material it’s not being used to identify fraud. It’s an interesting area for the National Crime Agency (NCA) and police forces, but not for us,” Denison says.
The one area technology does have an impact on is the sheer volume of material the SFO has to deal with. The SFO’s digital forensics team was dealing with about 2.5 million documents a month in April last year, and by the summer of 2017 this had gone up to five million a month, and by the end of the year it was 10 million a month.
“We don’t see that dropping at any rate – it’s likely to keep going up and that’s reflective of the type of material the company and individuals are generating and it’s a key challenge we’ll be facing,” Denison states.
Our cases are getting much bigger… now we have a case that has 50 million documents, and we’re getting to the point where you have to use technology to review this material
Data in general is an area the organisation is investing time and resources into, and not just in the front of house. On the back-office side of things, the organisation has an information programme running at the moment aiming to get the most out of non-case work data or non-evidential data.
“That’s looking at how we can generate that data, how we manage that data and, of course, making sure we’re compliant with GDPR and other legislation which we need,” Denison says.
This is one of several other IT projects that is ongoing, one of which includes looking at automating workflows.
At a time of increasing demands for the SFO, Denison – and the technology estate he oversees – are playing a part in making sure the organisation can cope with the scale and demand.
Auditors uncover £15m in fraud and errors and urge vigilance against dangers posed by pandemic
Joint Council for the Welfare of Immigrants files judicial review
EHRC also flags up need for more data on impact of remote hearings
Department recruits for individual to apply design principles to solve government challenges