The UK’s Serious Fraud Office (SFO) has an investigative role which is rather different from that of most regulators or law firms. It is investigator and prosecutor, taking on only seven to ten new cases each year, each of which might last between five and seven years.
The SFO has taken OpenText Axcelerate (which it acquired with the purchase of Recommind) to help it deal with the serious and complex fraud, bribery, and corruption cases which it manages, each involving huge volumes of documents and data which must be analysed to see if criminal activity has taken place and to run a prosecution where relevant. The Rolls-Royce fraud investigation which was settled in 2017 involved about 30 million documents. One case presently in hand has already turned up 50 million documents.
The OpenText press release is here. It quotes Ben Denison, Chief Technology Officer at the SFO, as saying that Axcelerate gives the SFO a way to “simplify its document review process, quickly capturing, categorising, and analysing data through a combination of advanced analytics features and built-in proprietary machine learning algorithms”.
Its benefit, Ben Denison says, is that Axcelerate
…can help significantly accelerate document review, including for legal professional privilege (LPP), saving the caseworkers valuable time and allowing them to focus on the most relevant and important content.
It means that “staff spent more time building a case instead of sifting through mountains of data”. The identification of privileged material is one of the hardest parts of an SFO investigation.
The Law Society Gazette goes for a catchy “Robo-lawyer” headline, focusing on the ability “to recognise patterns, group information by subject, organise timelines, remove duplicates and eventually be able to sift for relevance.”. The Gazette article draws attention to two aspects which matter to those involved in investigations – jobs and cost.
On jobs, the SFO says that its use of the new technology will not cost jobs but enable investigators to work through their cases more quickly. The SFO is not the first regulator I have seen saying this – everyone wants to do the job faster and at lower cost, but these imperatives are more exaggerated at a regulator. Regulators face criticism from the public, from government, and, indeed, from companies under investigation if the exercise takes too long, and they are criticised equally if seen either as too slow and heavy-handed or as insufficiently probing. The focus is less on reducing head-counts and more on swift and accurate results.
Ben Denison is quoted in the Gazette as saying that the technology will enable the SFO “to meet our disclosure obligations and deliver justice sooner, at significantly lower cost”.
Legal Futures focuses on a different aspect of the job losses, with the headline “Bad news for barristers: SFO adopts AI-powered document review after successful test in Rolls-Royce case”. The use of technology will, it says “put out of work barristers who [the SFO] previously used to identify material subject to legal professional privilege”.
One tweeting barrister’s reaction to that was relief at the thought that dealing with the SFO’s data (whether as prosecution or defence he did not say) would be much easier if it had been subjected to the kind of analysis which Axcelerate would bring.
This aspect often gets overlooked, in litigation as in regulatory investigations and prosecutions. Many eDiscovery / eDisclosure people see the job as done once inspection or production has been been provided to the other side. Other doors open as that one closes, however, and the hard legal work begins when the lawyers get their hands on the documents. The role for analytical tools like those in OpenText’s Axcelerate does not end with the conclusion of disclosure.