Beth Ashmead-Latham 12 Apr 2018 03:32pm

SFO’s new AI to “forever change” legal profession

In a landmark change, the Serious Fraud Office (SFO) will now use an automation tool to analyse legal documents

The UK financial watchdog says it will speed up investigations, at a lower cost and with fewer errors than lawyers – and that it comes at a time when using technology “is no longer optional”.

All new SFO cases will use the technology, beginning this month.

“The amount of data handled by our digital forensics team has quadrupled in the last year,” said the SFO’s chief technology officer, Ben Denison.

“[Technology] is essential given the volume of material we are dealing with and will help ensure we can continue to meet our disclosure obligations and deliver justice sooner, at significantly lower cost,” he added.

The SFO has already implemented a pilot robot which works at 2,000 times the speed of a lawyer. This was used during the Rolls-Royce investigation last year – the first UK criminal case to use AI – in which 30 million documents were submitted for review.

It will also begin using Axcelerate, an AI powered document review system from software company OpenText that recognises patterns, groups information by subject, organises timelines and removes duplicates.

Mark Barrenechea, OpenText’s vice chair and CEO, said that technology’s ability to review and analyse vast amounts of information will, “Forever change the way the legal profession operates”.

He added that the volume of material on current SFO cases already exceeds that reviewed in the Rolls-Royce case , with one listing 50 million documents requiring review and another larger than both cases combined.

Richard Anning, head of the IT Faculty at ICAEW, said, “We see the scanning of documents and other text files becoming more common in the work of accountants (such as scanning the contract estate to spot leases not declared), ‘turbo-charging’ what accountants can do. It is important to note that results are unlikely to be 100% accurate, they will need a sense-check to make sure they are valid (and where appropriate reflect human values), and human input may be necessary to decide what to do next,” he added.

Vijay Rathour, Grant Thornton partner and member of its Digital Forensics Group, welcomed the move saying, “Study after study shows that machine learning tools are far more capable at finding evidential records and excluding non-relevant material from consideration, than legacy ‘gold standard’ methodologies.”

However, he emphasised the importance of using these technologies to enhance rather than replace the human investigator role.